Mr Hughes was driving his camper van along a single-carriageway on a Sunday afternoon. He was driving faultlessly, and under the speed limit, but did not have a full driving licence and was not insured. He rounded a curve and saw a car speeding towards him, on the wrong side of the road. There was nothing Hughes could do to avoid a collision, and the crash killed Mr Dickinson, the driver of the other car. The crash was entirely Dickinson's fault, and evidence revealed that Dickinson had been extremely tired, and under the influence of heroin. Even so, the Court of Appeal ruled that Hughes could be convicted of causing the death of another person by driving a motor vehicle on a road without insurance and without a licence, contrary to s 3ZB of the Road Traffic Act 1988 (RTA 1988). After all, the offence required only that the defendant (1) cause the death of another person by driving a motor vehicle on a road; and (2) not be licensed or insured at the time of driving. Hughes was undeniably uninsured and unlicensed at the time of the accident. It was equally clear that had Hughes not been driving along at 45–55 mph on the single carriageway in the opposite direction to Dickinson at precisely the fateful time and place, Dickinson would not have collided with Hughes with the same force, and so would not have died in precisely the same way and at precisely the same time. Hughes’ driving was an undeniably proximate and significant cause of Dickinson's death, and there was no break in the chain of causation.
On appeal though, the UK Supreme Court reversed the Court of Appeal's ruling, holding that despite the fact that by rights, Hughes ought not to have been on the road at all, he could not, in any real sense, be said to have ‘cause[d]… death… by driving’, since there was nothing ‘in the manner of his driving which [was] open to proper criticism’.Footnote 1
Commentators have welcomed the liability outcome in this case – nobody seriously argues that convicting Hughes of a homicide offence on these facts would have been desirable. However, the court's logic has been less well received. Simester and Sullivan argue that the Supreme Court's argument erroneously conflated causation and culpability, and imply that since it is not within the Supreme Court's competence to adjudicate on ‘the propriety, in constitutional and human rights terms, of convicting non-culpable persons of serious criminal offences’, it ought to have applied the law literally and upheld Hughes’ conviction.Footnote 2 Writing separately, Simester added that the judgement in Hughes puts the legal understanding of causation ‘at risk of losing its shape’, and that ‘[a]ny discomfort about [convicting Hughes] should be directed at the legislator’.Footnote 3
But there is a better way of analysing the ruling in Hughes – one that does not put the concept of causation at risk of losing its shape, while simultaneously offering us the resources to exclude most innocuous conduct from the net of criminal liability. On this view, Hughes is not really a case about causation at all – it is about something prior to that: the conduct element of a criminal offence.
In this paper, I argue that instead of analysing a criminal offence's conduct element in terms of acts and omissions, we should ask whether the defendant has belied a contextually salient expectation as to how she should, or routinely would, conduct herself. Slightly different sets of expectations are salient depending on whether our interest is in questions of criminalisation, or the proper interpretation of existing offences, but the criminal law is normally interested only in conduct-tokensFootnote 4 that belie a relevant expectation in some way. Belying such an expectation need not itself suggests any culpability but it does mark out the conduct as remark-able, in the sense of being ‘worthy of remark’.
I defend this ‘Remark-able Conduct Requirement’ (RCR) as follows: after explaining where my argument sits within the broad framework of a criminal offence (Section 1), in Section 2, I build on the distinction, familiar within criminalisation theory, between true omissions and mere non-doings. I propose that, analogously, we should also distinguish between true commissions and mere doings. Combined, this generates the RCR for the purposes of criminalisation. I explain how the RCR narrows the field of conduct-tokens that are appropriate candidates for criminalisation. In Section 3, I explain how we can adapt the RCR for interpreting existing offences, and show that while doing so is not radically revisionist, it simplifies the application of the law, and helps us make sense of cases like Hughes. Most importantly though, it offers a principled argument to limit the scope of offences drafted in overly broad terms, by excluding innocuous doings from potential criminal liability, even when the statutory language used to specify an offence's conduct requirement appears capable of capturing them. Finally, in Section 4, I argue that the RCR also suggests ways to make progress on philosophical puzzles about how we should conduct ourselves.
1. Stage setting
Before embarking on this project, it is important to get a sense of its place within the broad framework of a criminal offence. In doctrinal criminal law, a person is liable to a criminal conviction if she is
(a) a responsible agent (ie she meets the law's capacity conditions of age, sanity, etc), who is
(b) capable of voluntary conduct (so, not in an automatic state), and is
(c) exercising that capacity for voluntary conduct (so, not involuntarily tripping, falling, or moving reflexively), to
(d) perform the offence's actus reus, with the
(e) necessary mens rea for it, and
(f) without any justification or excuse.
The actus reus – point (d) above – may include elements of conduct, consequence, and circumstance. This paper is about just the conduct element of the actus reus of an offence. It discusses the role that this tiny but essential cog plays at two stages in substantive criminal law: first, when deciding what sorts of conduct-tokens are normatively appropriate candidates for criminalisation; and secondly, when interpreting the actus rei of existing offences. As such, my arguments are addressed to two audiences: the first part is addressed to conscientious legislators, and the second, to those who interpret and apply existing criminal laws.
Conventionally, the conduct element of a criminal offence is fleshed out in terms of acts and omissions. While I think that this is a misstep, I cannot defend that proposition here. Instead, in this paper I propose the RCR as a plausible alternative way of understanding the criminal law's conduct requirement, and show how it outperforms the conventional approach. While this would not, in itself, prove the irrelevance of the Act-Omission Distinction (AOD), it would show that the criminal law can survive and thrive without the AOD.
2. Remark-able conduct and criminalisation
Although for the most part in this paper I propose to sidestep the AOD, the literature on it does yield some valuable insights about the sorts of conduct that a conscientious legislator would not ordinarily support criminalising. These insights, in turn, have implications for the interpretation and application of existing criminal offences.
Theorists on both sides of the debate about the significance of the AOD in criminalisation decisions agree that in ordinary language, not all non-doings are omissions.Footnote 5 Even those who argue that omissions are as plausible as acts qua candidates for criminalisation, do not extend their claim to mere non-doings. While parliament can, of course, designate a mere non-doing as the conduct element for an offence, nobody argues that a general policy of doing so is normatively appealing. Usually, we should only consider criminalising ‘true’ omissions, though in itself, being a true omission generates no positive argument for criminalisation.
(a) Omissions
So how does one distinguish ‘true’ omissions from ‘mere’ non-doings in ordinary language? Kleinig,Footnote 6 who relies extensively on Feinberg'sFootnote 7 analysis, offers this answer:
‘S omitted to do α if and only if:
(1) S did not do α;
(2) S had the opportunity to do α;
(3) S had the ability to do α;
(4) S had good reason to believe that he had the opportunity and ability to do α; and
(5) α was reasonably expected of S, because
(a) S, or those in S's position, ordinarily do α; or
(b) S had a responsibility to do α; or
(c) S was obligated to do α; or
(d) α was in some other way morally required of s’.
Conditions (2) to (4) above refer to the agent's practical capacity to do the thing she allegedly omits to do. They explain why I do not ‘omit’ when I fail to orbit Neptune, or fail to switch on a light that I don't realise is sound-activated. These conditions carry over into the criminal law. Even if we assume that my non-doings in these cases can be called ‘conduct’, there is a real sense in which they are non-voluntaryFootnote 8 and therefore incapable of satisfying the criminal law's voluntary conduct requirement. Note that this leaves open the possibility of inadvertent omissions.Footnote 9 One can have the practical capacity to do something while absent-mindedly failing to exercise that capacity, and such a non-doing will be an omission, if condition (5) is satisfied.
Condition (5) – the ‘reasonable expectation’ requirement – reflects the fact that in ordinary usage, the term ‘omission’ picks out non-doings that are ‘remark-able’, in the sense of being ‘worthy of remark’. They, in some way, belie an expectation. The expectation that is belied may be legal, contractual, or moral, but it need not be – even a statistical expectation, or an expectation based on an established, but not mandated, routine would do.Footnote 10 Kleinig explains that these expectations are ‘conventional manifestations of a social milieu’,Footnote 11 and notes that they may depend on one's role responsibilities.Footnote 12 In sum, for Kleinig, the term ‘omission’ is not necessarily a pejorative designationFootnote 13 – it picks out note-worthy, but not necessarily even prima facie blameworthy non-doings. When, unusually for her, Sara fails to fetch her newspaper from the garden in the morning, we can say that she omitted to bring in the newspaper,Footnote 14 since a neighbour, noticing this anomaly, may well think it worthy of remark. But in the absence of that regular course of conduct, it would be strange to call this an omission – the neighbour would have no anomaly about which to remark. Similarly, if I usually walk to work, but one day I don't, it makes sense to say that I omitted to walk to work. But if I have no usual way of getting to work, or if I usually cycle to work, then not walking to work could hardly be called an omission. It would not be a deviation from what I might be ‘reasonably expected’ (in Kleinig sense) to do. Note also that merely having a reason to act generates no expectation that I act – this is why I do not ‘omit’ to have dessert when I decide not to have it, even though my craving gives me good reason to have it.Footnote 15 But when I miss an appointment, whether it be my own regular swim, or a staff meeting, I belie a ‘reasonable expectation’ of me, and so I do omit to keep the appointment.
One may plausibly contest some details of Kleinig's distillation of the ordinary sense of an ‘omission’, but for this paper, I propose to adopt it. Provided that the reader accepts that omissions are anomalous non-doings, and that generally, only such non-doings are normatively appropriate conduct-tokens to consider criminalising, the details of what makes some non-doings anomalous do not affect my central claims.
Still, some elements of Kleinig's argument need further clarification. For instance, when Kleinig refers to reasonable expectations, he does not explicitly specify whose expectations he means. From the examples he offers (many of which are similar to those offered here), it is clear that Kleinig is interested not so much in the agent's own subjective expectations, as in the objective expectations of a hypothetical onlooker. But we still need to know what information to attribute to this onlooker. Kleinig's enumeration of the expectations suggests that the onlooker knows the moral norms and routines prevalent in the agent's society, and the agent's own responsibilities and routines. But what if these expectations conflict with each other? Consider this case: it is not considered polite to ask someone how much they earn, especially the moment one meets them. Amina, however, always does so. But on this occasion, when Amina meets Karim for the first time, she does not ask about Karim's income. Would it be correct to say that Amina omitted to ask Karim about his income? Kleinig's view offers no answer, but we can tweak it to provide one.
What qualifies as an omission depends on the context in which the question is asked.Footnote 16 The same conduct-token can be an omission in one context – Amina's friends, knowing how Amina normally behaves, might themselves find Amina's unexpected restraint remark-able – and a mere non-doing in another – they would still be surprised if a casual onlooker pressed them on Amina's incuriosity, since by reference to ordinary societal expectations, that is entirely mundane. This is because, although Kleinig does not make this clear, condition (5) above refers to expectations that are contextually salient. Kleinig is interested in whether a non-doing is an omission in the context of an argument about whether it is more appropriate to focus our criminalisation efforts on acts or omissions. In that context, the expectations of the ordinary member of society will be more salient than those of Amina's inner circle. The criminal law does not operate with the sort of granularity that would make the expectations of Amina's intimates salient, but on any plausible normative theory of criminalisation, societal expectations play an important role in guiding thinking about what should be criminalised. On occasion, when an agent has special role-responsibilities, the expectations of the sub-community of persons occupying the same role are also relevant. Our ordinary member of the society will be well-versed with its moral and routine-based expectations, as well as those of any relevant sub-community of persons occupying the same roles as the agent. She would find deviations from those expectations remark-able. Non-deviations though, whether owing to compliance with the expectations, or because there are no relevant settled expectations, would be unremark-able in the relevant sense.
This explains why, while her neighbour may rightly call Sara's exceptional (by reference to Sara's own personal routines) failure to fetch her newspaper in the morning an omission, it is not an omission for the purposes of the criminal law. But if most members of a society routinely fetched their respective newspapers before 9:00am, Sara's failure to do so might well be worthy of remark, even though there is no pejorative connotation to not bringing in one's newspaper.
Notice that making our conception of criminal law-related omissions sensitive to societal morality and routines means that different liberal societies, with different such expectations, will identify different non-doings as omissions. But in each, an omission will be a deviation from a contextually salient expectation to act.
Further clarification is also necessary to address cases in which the agent is pulled in different directions by conflicting considerations. For instance, if I was supposed to deliver a lecture, but did not, you might think that I omitted to do so. But what if I chose not to deliver the lecture because I was ill?Footnote 17 Should we revise our view and say that on balance, I was not expected to deliver the lecture, and so deny that I omitted to do so? Or should we maintain that I omitted to deliver the lecture, but say that I can offer an explanation that deflects any blaming assessment? In the context of Kleinig's broader project of studying the appropriateness of criminalising omissive conduct-tokens, we may rephrase the question thus: does my illness let me deny performing the conduct element of the actus reus of the hypothetical offence of omitting to deliver my lecture, or offer me a rationale-based defence to a blaming judgment flowing from satisfying, inter alia, the conduct element? The former view would require us to refer to something like ‘net expectations’ when identifying omissions. But this view is implausible – it would mean that rationale-based defences were inapplicable (because unnecessary) to all omissive crimes, though they remain relevant to commissive crimes. As far as I know, nobody defends that position.
If the latter claim is correct though, then at least when identifying omissions in the criminal law context, we need some way to distinguish considerations relevant to an expectation to act, from those that set up rationale-based defences for omitting to act. My preferred way of doing that is to distinguish between the considerations relevant to the domains of prima facie offences and rationale-based defences.
In previous work, based on an analysis of how conduct rules and decision rules typically slot into criminal law, I have argued that ‘offence definitions typically track the exceptions to the default position of universal unfettered freedom to [conduct oneself] as one pleases’ whereas defences track the permissive exceptions to those exceptions. In other words, in modern liberal systems of criminal law, offences track expectations that narrow our liberties, whereas defences track the permissions that are the exceptions to these narrowings.Footnote 18 On this view, the imperative liberty-limiting expectations (ie expectations that we perform an act, or avoid performing an act) are relevant to offence stipulations, whereas exceptional liberty-expanding permissions (ie permissions to do otherwise prohibited acts, or to omit otherwise required acts) are relevant to rationale-based defences. Accordingly, in my example, my imperative liberty-limiting expectation was that I perform an act (deliver the lecture). Therefore, when I did not, I omitted to deliver the lecture. However, an exceptional liberty-expanding permission applied to me (I am permitted to rest if I feel unwell). Therefore, I had a rationale-based defence to offer in respect of my omission.
This approach to classifying expectations as to conduct leads us to an austere (but I think, defensible) view of the considerations relevant to identifying true omissions for criminalisation-related inquiries. Only considerations that imperatively limit our default unfettered freedom to conduct ourselves as we please are relevant when identifying contextually salient expectations. Factors that permissively expand (by carving out exceptions to imperative limitations) our freedom to behave as we please, relate to rationale-based defences rather than to the conduct component of the actus reus of a prima facie offence. They must therefore be excluded.
Consider how this shapes our analysis in the classic Trolley Problem.Footnote 19 When examining the role of the AOD, we should exclude from consideration factors that support only rationale-based explanations of why the agent cannot be blamed for committing a prima facie offence (ie actus reus + mens rea) by way of her chosen conduct-token. These might include factors like the numbers of victims on either fork of the tracks, their ages, their histories, the families that might be affected by their respective deaths, and the costs (in terms of time, resources and risks) to the agent attendant with each alternative course of conduct. Such factors may render the prima facie criminal conduct-token exceptionally permissible, but they do not suggest the absence of an expectation that she conduct herself differently. Factors that limit the agent's default freedom to conduct herself as she pleases are, however, relevant. Hence the fact that each track has (at least) one person on it may generate expectations that the agent divert the trolley, and simultaneously, not divert it. In such a case, the agent's voluntary choice of either course of action would satisfy the criminal law's conduct requirement, and call for an explanation. A successful explanation would avert any blaming judgment flowing from the agent's chosen conduct-token.
In sum then, only non-doings that
(a) are ‘performed’ by persons exercising their practical capacity for voluntary control, and
(b) belie a contextually salient (ie stemming from societal morality or routine) expectation
are ‘true’ omissions. In general, only this subset of non-doings are even potentially of interest to conscientious legislators considering questions of criminalisation.
(b) Commissions
The claim that true omissions are a subset of non-doings is fairly well trodden theoretical ground. But it hints at an intriguing possibility – one that, as far as I can see, has not yet been seriously explored.Footnote 20 The same arguments that suggest that not all non-doings are omissions, might also suggest that not all doings are ‘acts’. Accordingly, if mere non-doings are not appropriate candidates for criminalisation, then perhaps neither are ‘mere’ doings. I want to take that possibility seriously here.
To do that, we must temporarily set aside the term ‘acts’ – that term is in ordinary usage, and is used in many senses, and so our intuitions about it may be difficult to refine. To investigate whether, as for non-doings, only a subset of doings interests the criminal law, we would be better advised to follow Honoré's lead, and use the word ‘commissions’ instead.Footnote 21
Here are some things I do: I age, I shed skin cells and hair, I sweat, I breathe and blink at a regular pace, and I unthinkingly drum my fingers on the table as I read.Footnote 22 Each morning, I wake up, I reach for my spectacles, and soon after, I brush my teeth, comb my hair, and have a cup of tea. Occasionally, I yawn in public without covering my mouth, or cut across my neighbour's lawn on my way to the shops. Suppose we tried to identify ‘commissions’ – the subset of doings of interest to the criminal law – in the same way as we identified omissions, which of these doings would make the cut?
Recall that omissions are those non-doings of persons exercising their practical capacity for voluntary control that belie a contextually salient expectation. So our posited set of commissions would be those doings of persons exercising their practical capacity for voluntary control that belie a similar expectation. Let us apply this test.
Since it is beyond my practical capacity to avoid doing things like aging, and shedding epithelial cells and hair, these doings are non-voluntary. Clearly, they are not the sorts of doings in which the criminal law is interested.Footnote 23 I can control, to some extent, how much I sweat, and how often I breathe and blink. I have even more control over my unthinking physical tics – I can avoid drumming my fingers on the table if I only thought to. However, there is no relevant expectation that I not breathe or blink at my normal rate, or drum my fingers on the table. Therefore, these doings should also not count as commissions. This tallies with our experience of the criminal law – my sweating or breathing or blinking at my normal (unreflective) rate, or absent-mindedly drumming my fingers on the table, are not doings that the criminal law usually considers criminalising.
I have still more control over the various doings that form part of my daily morning routine, but again, since there is no relevant expectation that I not do those things, none of them is worthy of remark. Analogy with the omissions arguments would suggest that these things too are not commissions, and they would not be commissions even if, for instance, I had reason to abstain from doing any of these things.Footnote 24 So for instance, my drinking tea would not be a commission even if I drink too much tea, and so have reason to cut down. But how does this conclusion compare to our experience of the criminal law? Well, ordinarily, these are not the sorts of things that we consider criminalising. One could certainly add facts to make these routine doings more plausible candidates for criminalisation.Footnote 25 Perhaps when D drums her fingers on the table, she activates a concealed switch that electrocutes V, or conveys the green signal for an accomplice to commit a burglary. Would that make D's seemingly quotidian conduct-token remarkable?
I think that because of the context of our inquiries – our ultimate interest in issues of criminalisation – the answer depends on what D understands about her conduct. We want to be able to draw particularised conclusions about D from, inter alia, D's exercise of her moral agency to perform a conduct-token. The conclusions that we can draw about D from her exercises of moral agency to perform unremarkable conduct are limited – in those exercises of moral agency, D shows herself to be like a generic other person. Only when D exercises her moral agency to do something anomalous can we draw particularised conclusions about D. This makes D's own understanding of the conduct-token she is performing more salient than other descriptions of it – after all, it is by reference to that understanding that she exercises her moral agency. For D's conduct to be a commission, she should, at a minimum, realise that the conduct-token that she is exercising her moral agency to perform, is remark-able. She need not know that the switch will electrocute V, or the content of the secret message being conveyed by her conduct – those matters are relevant, but only to her culpability (and in the criminal law context, her mens rea and some rationale-based defences). But even if she knows only that her conduct carries some special significance, be it activating a switch, or conveying some secret message, ie that it is not merely drumming her fingers on the table, it is appropriate to say that her exercise of moral agency is remark-able and of potential value in drawing particularised conclusions about D. It is, therefore, a plausible candidate conduct-token for criminalisation. Conversely, if she does not realise that her conduct-token is anomalous, prima facie her exercise of moral agency is uninteresting to the criminal law. This is also how the criminal law treats such cases. If T secretly rigged the table so that D's drumming on it caused V to be electrocuted, D would be treated as T's innocent agent, and exonerated, not merely because she lacked the mens rea for an offence, but because she did not, in the relevant sense, act as an effective autonomous agent at all. T would be treated as the principal in relation to electrocuting V, and as far as the criminal law is concerned, D would drop entirely out of the picture qua agent.
When I yawn in public without covering my mouth, I belie a relevant expectation of me. Such behaviour is considered impolite, and so it belies the societal expectation that I will behave politely. It is an anomaly, and so it is remark-able. The same is true when I cut across my neighbour's lawn – in fact, the relevant societal expectation is also the basis for the tort of trespass. Therefore, by virtue of belying that expectation, what I do is remark-able. These conduct-tokens may properly be thought of as being commissions, and so they are plausible candidates for criminalisation. There will be little reason to criminalise most commissions. However, given an institutional appetite for criminalisation, they are conduct-tokens upon which criminal liability can plausibly be founded, provided that various other requirements – responsible agency, consequences, circumstances, mens rea, an absence of defences, etc – are met. Similarly, if we swap my tea-drinking for smoking, and add contextual facts such as my having an infant at home, then given the parental role I occupy, society may well frown upon me smoking at home. In this new context, my smoking would belie a societal expectation and be the sort of remark-able conduct which we might plausibly consider criminalising.
But just as was true for omissions, the mere fact that a doing is a commission offers no positive support for its criminalisation. The institutional appetite for criminalising the commission depends on whether we have reason to discourage the concerned commission, and on contingent systemic factors including the existence of the resources and political will to use the criminal law in this way, and the anticipated consequences of criminalising the conduct. And even criminalised commissions only generate liability if other requirements like responsible agency, mens rea, an absence of defences, etc are met. But doings that are not commissions don't even get this far – they are normatively unappealing candidates for criminalisation. To be clear, it is within parliament's gift to nevertheless criminalise mere doings despite these normative arguments against doing so. Perhaps there are countervailing normative arguments in support of criminalising mere doings – perhaps the underlying societal expectations ought to change, and legislators would like to use the law to effect this change.Footnote 26 But in the absence of such reasons, we would expect such instances of criminalisation to be subjected to normative criticism.
It seems plausible then to think of ‘acts’ as ‘commissions’ in the sense described, and thereby to distinguish between doings that are plausible candidates for criminalisation, and doings that are normatively unappealing ones. When we combine this idea with the distinction between omissions and non-doings previously discussed, we see that, prima facie, conduct-tokens are normatively implausible candidates for criminalisation unless they:
(a) engage the agent's practical capacity for voluntary control, and
(b) belie a contextually salient expectation.
Crucially, it does not matter whether the conduct-token is a commission or an omission, so long as it belies a contextually salient expectation as to conduct. By belying such expectation, the conduct-token becomes ‘remark-able’, ie worthy of remark, and a plausible candidate for criminalisation. But (unless countervailing considerations apply) a conscientious legislator should find unremarkable conduct-tokens normatively unappealing candidates for criminalisation.
3. Remark-able conduct and existing offences
When we move from interpreting ordinary language with a view to answering questions about criminalisation, to interpreting legal language with a view to interpreting existing criminal offences, we can expect some changes to the foregoing analysis. For one thing, as the context changes, what is contextually salient changes. Since we are interpreting a particular offence, we might expect the terms of that offence to provide definitive guidance as to what conduct is expected of people to whom the offence applies. That does, in fact, happen on occasion. Consider s 183 of the Companies Act 2006. The offence applies to directors of companies, and it is committed when a director fails to declare an interest in an existing transaction or arrangement as required by s 182 of the same statute. The conduct expected of the director in respect of this offence emerges clearly from the words of the offence, read with related provisions of the statute: she is expected to declare any interest in an existing transaction or arrangement covered by s 182.
In fact, many offences criminalise the performance of narrowly specified conduct-tokens in particular circumstances, and/or with particular consequences. These conduct-tokens may be omissions (eg to perform specified duties like paying taxes or reporting certain crimes) or commissions (eg penetrating a person's vagina, anus, or mouth with one's penis; or driving in a specified manner – be it dangerously, carelessly, or inconsiderately; or driving at all when unlicensed or uninsured). Only conduct that belies the expectations salient to such an offence is a plausible candidate for liability under the offence. Other expectations are irrelevant.
But we also have offences that specify the salient expectations as to conduct only in very broad terms. In such cases, we need to look elsewhere for more detail. Consider the offence of gross negligence manslaughter. This offence is committed when a person owes a duty of care to another, and causes the latter's death by breaching that duty in a manner that is both grossly negligent, and creates an obvious and serious risk of death.Footnote 27 While the conduct-token of interest for this offence is ‘breaching a duty of care’, that is hardly specific enough as to the expectations applicable to each potential defendant. Therefore, in deciding different cases, we must look beyond the elements of the offence. Given that a criminal offence is not a standalone rule, but is part of an entire framework of rules governing and facilitating our lives, the most obvious place to look is elsewhere in the law. This too is, in fact, what we do. So, when the defendant is a medical doctor alleged to have caused (whether by act or omission) the death of a patient through her medical negligence, we often refer to medical law to identify the expectations applicable to the defendant.Footnote 28
But what if the law does not articulate any expectations as to conduct? At this point, we need to go further afield. In developed liberal legal systems, we would expect the legislator to have been conscientious (or at least, we presume that they had been conscientious) when legislating. Therefore, it makes sense, in the absence of more specific guidance, to read conduct stipulations as referring to conduct-tokens that would be remark-able for the conscientious legislator: ie conduct-tokens that belie a societal expectation, including, where applicable, expectations stemming from the role(s) that the agent was essaying. The conduct of an agent cannot plausibly be said to satisfy the conduct element of an offence if it belies no societal expectation. But if it does, we have a pro tanto (though not conclusive) reason to think that the conduct-token satisfies the offence's conduct element. Again, this is an accurate description of what the criminal law in fact does. Where the agent is alleged to have committed an offence acting in a responsible capacity, a failure to live up to the expectations attendant to the agent's role is a qualifying conduct-token. Accordingly, legal recognition has been granted to expectations to act arising from the societal norms governing settled relationships of dependenceFootnote 29 and voluntarily assumed obligationsFootnote 30 (including contracts).Footnote 31 Where no special role responsibilities apply, a failure to conform to societal norms will suffice. Thus, the finding in R v Miller Footnote 32 that a person who accidentally created the risk of a fire ought to have taken steps to address that risk essentially recognised the societal norm requiring one to ‘sort out the messes one creates’. But although we would arguably agree that morally, we should help someone facing a serious risk to life, especially when it is easy and not costly to do so, English criminal law chooses not to treat the failure to make an easy rescue as conduct that satisfies the actus reus of a general homicide offence. Perhaps in this case, the pro tanto reason to treat a failure to make an easy rescue as qualifying conduct is overridden.Footnote 33
So far, it seems like interpreting criminal offences using the RCR makes little change to the law. But recall that the RCR would apply not only when determining whether non-doings are qualifying omissions, but also when determining whether doings are qualifying commissions. This latter application of the RCR would significantly change the analysis of acts, since not all doings would automatically be considered commissions capable of meeting an act requirement in an offence. Qualifying commissions too would need to belie the same sorts of expectations that qualifying omissions belie.
The idea that some doings are not qualifying commissions even for offences with ‘any conduct’ or ‘any act’ conduct elements, is novel in English law but has a foothold in civil law systems. In German law, quotidian or ‘neutral’ acts that in fact contribute to a crime do not generally satisfy the offence's conduct element, though the German tests for identifying acts that are ‘neutral’ differ from the RCR.Footnote 34 Moreover, even those writing in the common law tradition recognise that causal chains attach to events that deviate from the normal – that belie expectations. Consider this from Simester:
… in the law… the ascription of causal responsibility is a normative as well as a mechanical issue… it seems acceptable to say that Pete's not smoking… caused his health to improve, or that the railway gatekeeper's failure in Pittwood to close the gate before the train came through caused the resulting accident. By contrast, it would be misleading to say that the train's arrival caused the accident, although that was also a causal factor. There are inevitably multiple ingredients in the occurrence of an event: We tend to reserve the language of causation for those which are the most salient to our explanations.Footnote 35 (Internal references omitted)
Feinberg defends the same proposition in detail, arguing that when we seek an explanatory ‘causal citation’ for some surprising or unusual event in order to ascribe responsibility to a moral agent, what we are interested in is agential conduct that is both: a causal condition; and a deviation from what might have been expected. Accordingly, we ask, ‘What unusual cause [in this context, what unusual voluntary human conduct] accounts for this unusual result?’.Footnote 36 Similar arguments are also made by others.Footnote 37
Notice that while these arguments are couched in the language of causation, what they actually refer to – the ‘causal citation’ – is the element at the beginning of the causal chain. For criminal law related purposes, as the RCR recognises, this is the conduct-token. Quotidian conduct-tokens are usually uninteresting to the criminal law; its main interest is in unusual conduct-tokens. Thus, in Simester's train accident example, the train driver performs conduct in bringing the train to the junction, but since this conduct is not anomalous, ie ‘remark-able’, it is not an appropriate conduct-token to which to attach a causal chain for criminal law purposes.
(a) The net of criminal liability
How would this development of the RCR for interpreting offences work in practice? Well, as I have pointed out, adopting the RCR will not have a radically expansionist effect on liability outcomes. For instance, the RCR makes no difference whatsoever to the scope of liability in offences that criminalise the performance of narrowly specified conduct-tokens. It does not matter whether these conduct-tokens are omissions (eg to perform specified duties like paying taxes, or reporting certain crimes)Footnote 38 or commissions (eg penetrating a person's vagina, anus, or mouth with one's penis; or driving in a specified manner). That said, adopting the RCR will sometimes make such cases easier to understand.
To see how, let us revisit Hughes. Recall that Hughes was charged with an offence under s 3ZB of the RTA 1988: causing death by driving while unlicensed or uninsured. The Supreme Court held that since the offence required causing death by driving, there needed to be something objectionable in the manner of driving – ‘some act or omission in the control of the car, which involves some element of fault, whether amounting to careless/inconsiderate driving or not’, which ‘contribute[d] in some more than minimal way to the death’.Footnote 39 In other words, the Supreme Court held that driving per se was not the conduct element of this offence – it was objectionable driving. Effectively, it ruled that when parliament referred to people driving while unlicensed or uninsured, it was not setting out the conduct element of the offence, but instead identifying persons to whom the offence applied – those eligible to commit the offence. The conduct element of the offence was a deviation from the relevant set of contextually salient expectations applicable to eligible persons. The Supreme Court interpreted s 3ZB as stipulating that the contextually salient expectations related to the manner of driving – in respect of this particular offence, a person eligible to commit the offence was expected to drive faultlessly. Accordingly, the conduct element of the offence was driving in a manner that belied that expectation. It could be performed either by act, ie by doing something one was expected not to (eg marginally exceeding the speed limit), or by omission, ie by not doing something that one was expected to (eg not ensuring that each tyre was adequately inflated). If that conduct could be linked to the specified consequence – the death of another – using ordinary rules of causation, then prima facie, the driver committed an offence. Hughes did not belie any contextually salient expectation applicable to him as a person eligible to commit an offence under s 3ZB – he drove faultlessly. Hence, he did not engage in any conduct of interest in relation to this offence. The causation analysis, which links conduct to consequence, never took off, since there was no relevant conduct to which it could attach.
We may query the Supreme Court's conclusion that parliament had made objectionable driving, rather than just driving, the conduct element of the s 3ZB offence, but there is something to be said for it. If we compare s 3ZB with the other offences in the same statute that s 3ZB mentions, viz driving while unlicensed (s 87) and driving while improperly insured (s 143), we see some striking differences. While s 3ZB speaks of causing death by driving, s 87 says that it is ‘… an offence for a person to drive on a road a motor vehicle… otherwise than in accordance with a licence…’, and s 143 says that ‘a person must not use a motor vehicle… unless there is in force in relation to the use of the vehicle by that person… a policy of insurance’ (emphasis added). These latter offences clearly criminalise the very act of driving (or using) a motor vehicle. For these offences, the contextually salient expectation which the defendant must belie to be of interest, is that she not drive/use the motor vehicle.
Now consider s 1 of the RTA 1988: ‘A person who causes the death of another… by driving a… vehicle dangerously on a road… is guilty of an offence’ (emphasis supplied), and s 2B of the RTA 1988: ‘A person who causes the death of another… by driving a… vehicle on a road… without due care and attention, or without reasonable consideration for other persons… is guilty of an offence’ (emphasis supplied). These offences, like s 3ZB, use the phrase ‘by driving’. The contextually salient expectations belied by persons who commit these offences clearly relate to the manner – dangerously, without due care and attention, or inconsiderately – rather than the fact of driving. In other words, the contextually salient expectation in relation to these offences isn't that the defendant not drive – it is that she not drive in a particular manner – dangerously, carelessly, or inconsiderately as the case may be. The adverbs ‘dangerously’, ‘carelessly’ or ‘inconsiderately’, specify how seriously anomalous the manner of driving must be to meet the conduct requirement of these offences, but even where no such adverbs are used (as in s 3ZB), the phrase ‘by driving’ suggests that the context-appropriate background expectations relate to the manner, rather than the fact, of driving.
Given s 3ZB's structural similarities with provisions like ss 1 and 2B of the RTA 1988 (and other similar provisions in that Act); and structural dissimilarities with provisions like ss 87 and 143 of the RTA 1988, the Supreme Court's reading of s 3ZB is at least plausible.
But even if we disagree with this doctrinal argument about the interpretation of s 3ZB of the RTA 1988, the theoretical point stands. If we accept that only some doings – those that belie a contextually salient expectation – are of interest to the criminal law in relation to an offence, then in cases like Hughes we should ask whether the defendant performed any conduct-token that qualified for the purposes of the concerned offence. One benefit of thinking of an offence's conduct element in terms of the RCR rather than acts and omissions is that it forces us to identify the set of contextually salient expectations relevant to the particular offence, by reference to which the significance of a defendant's conduct-token is to be assessed. In doing so, a Hughes-style avenue of argumentation is opened to us.
Occasionally, some offences will specify the necessary conduct, but use terms that are equivocal as to whether a commission or an omission is required. Thus, the offence of theft can only be committed if the defendant ‘appropriates’ (property belonging to another). Nothing in the term ‘appropriates’ demands an act, though English courts have interpreted it as though it does.Footnote 40 Similarly, English courts have interpreted the common law offence of constructive manslaughter to require that the underlying offence be committed by act rather than omissionFootnote 41 although nothing inherent in constructive manslaughter requires that reading. Since the RCR is an alternative to adopting the AOD, under it, we would have a pro tanto reason to think that such offences could also be committed by omission.
But notice that this pro tanto reason can be overridden – and in the context of both the foregoing examples, perhaps it is. In the law of theft, the elements of appropriation and dishonesty have been interpreted extremely widely, thus vastly expanding the net of liability. Against that backdrop, we arguably have good policy reasons to counterbalance that expansiveness by, say, insisting on appropriation by positive act. Similarly, given the concerns about the unfairness inherent to constructive criminal liability, especially for homicide offences,Footnote 42 we may have good policy reasons to limit constructive manslaughter by insisting that the underlying offence be committed by act, rather than omission. To be clear, it is plausible to respond that tweaking the conduct requirements of the theft and constructive manslaughter offences is an inapt solution to deeper problems with these offences. My claim here is simply that it is possible to think of reasons that might override the pro tanto reasons generated by the RCR.
If these pro tanto reasons are not overridden, the RCR may well have some expansive effect on the law, but arguably, it would be normatively appropriate. Indeed, commentators already express some puzzlement as to why, normatively, constructive manslaughter cannot build on an omissive base crime,Footnote 43 and theft cannot be committed by passive appropriation.Footnote 44
There are many other offences for which the performance of any conduct or a very broadly defined set of conduct-tokens leading to a specified consequence, or preparatory to future criminality, will do. Such offences include murder, assault, battery, offences under ss 18, 20, and 47 of the Offences Against the Person Act 1861, inchoate offences like criminal attempts under s 1 of the Criminal Attempts Act 1981, the ‘encouraging or assisting’ offences under ss 44, 45 and 46 of the Serious Crime Act 2007, ‘possession of articles for terrorist purposes’ and ‘collection of information useful for terrorism’ under ss 57 and 58 of the Terrorism Act 2000, and the ‘preparation of terrorist acts’ offence under s 5 of the Terrorism Act 2006.
Even in relation to these ‘broad conduct offences’, the RCR will not radically expand criminal liability. Partly, that is because the RCR applies only to the offence's conduct element, which is just part of the actus reus of the prima facie offence. As such a small cog in the overall machinery that generates a conviction, its expansive effect is naturally limited – any expansion in the range of qualifying conduct-tokens might well be cancelled out because the defendant lacked responsible agency, or did not satisfy some applicable circumstance or consequence elements of the actus reus, or lacked the mens rea for the offence, or had a rationale-based defence available.
But there is another factor that limits the expansionist effect of the RCR on broad conduct offences. Presently, many broad conduct offences can be committed by omission, provided that the omission breached a legally recognised duty to act. As I have demonstrated, the duties to act that the law recognises usually mirror the expectations to act which matter when applying the RCR to existing offences. These include expectations to act arising from other parts of the legal system, and in their absence, societal expectations (including, where appropriate, the expectations of people with the same role-responsibilities as the agent). Thus, the qualifying omissive conduct-tokens under the RCR are essentially the same as under existing law.
In relation to acts-based liability for broad conduct offences though, we might expect a difference. The RCR would mean that all doings would no longer automatically be of even potential interest in relation to these offences. Only commissions – acts performed by an agent exercising her capacity for voluntary conduct that belied a contextually salient expectation – would satisfy such offences’ conduct requirements. But far from expanding the net of criminal liability, this would contract it, since qualifying conduct is a necessary element of these offences.
Even in this respect, though, the effect of the RCR should not be overstated. Although mere doings have not hitherto been systematically excluded from the set of qualifying conduct-tokens for broad conduct offences, we have rarely noticed a problem since the criminal law has always been administered with a healthy dose of common sense. So far, therefore, we have hardly had to worry about mere doings being the basis of criminal charges.
But there are two reasons not to be comfortable with the status quo. First, occasionally common sense runs out. This threatened to happen all the way up to the Supreme Court in Hughes Footnote 45 and it did happen in a previous case with similar facts, in which the defendant was convicted.Footnote 46
And secondly, a recent proliferation of broad conduct offencesFootnote 47 has coincided with a growing realisation that, at least in the context of terrorism, the heavy-handed use of the criminal law is a greater worry. This has led commentators to worry that even innocuous acts like putting sugar in a cup of tea,Footnote 48 or writing a letter,Footnote 49 or eating porridgeFootnote 50 seemingly satisfy the conduct requirements of very serious crimes, with only mens rea and defence stipulations precluding convictions. These safeguards are not always adequate.Footnote 51 Although common sense has ensured that such conduct-tokens have not hitherto been made the basis for such charges, common sense is not enforceable. The fear of the over-zealous (and potentially discriminatory) use of such offences where the defendant is suspected of links to terrorism, has led commentators to explore various strategies to rein in the scope of such offences. Cornford, for instance, suggests that adopting a purposive approach to interpreting statutes might give us some of the resources needed, but concedes that ‘this approach requires courts to address difficult moral questions about the convictions and punishments that they are authorizing – questions which, ultimately, they may wish to avoid’.Footnote 52 Alternatively, if we read Hughes as building culpability into the rules of causation,Footnote 53 we could (at significant, probably prohibitive, cost to the coherence of the rules of causation) narrow the scope of broad conduct offences where the rules of causation are relevant – ie where some consequence must have been caused. But that fails to address the problem in respect of problematic broad conduct offences without consequence elements. By contrast, the RCR applies to any crime that requires the performance of some conduct, whether or not a consequence is required. It narrows the scope of all broad conduct elements.
The RCR, then, gives us the argumentative resources to deny that innocuous doings satisfy these offences’ conduct elements. If it were built into our legal understanding of the conduct element of a crime (like we currently build in the doctrinal tests for qualifying omissions), defendants would have a strong legal argument against prosecutions that fly in the face of common sense.
Of course, parliament could still legislate to make innocuous acts the basis for criminal consequences, but it would have to do by explicitly identifying the innocuous tokens in narrow specified conduct offencesFootnote 54 – broad conduct offences would not suffice.
In sum, then, the RCR offers an important argumentative bulwark against the over-expansion of the net of criminal liability for doings, while occasionally contributing to making badly motivated omissions eligible for criminalisation. In the most part though, its adoption would have little impact on the current net of criminal liability. It would not, therefore, be radically revisionist. This is because, even though the RCR cuts across the AOD in a way that suggests that the AOD is beside the point, it remains true that in today's liberal societies, we usually belie expectations by acts rather than by omissions. So, if I am correct about the RCR, then while the AOD is a good proxy for identifying conduct of interest to the criminal law, it is only a proxy. It carries no intrinsic weight in deciding what conduct is of interest to the criminal law.Footnote 55 It gives us prima facie, but not pro tanto, reason to think that acts are of interest whereas omissions are not.Footnote 56 And if the AOD is just a proxy for what is really important to the criminal law, it is replaceable. We can swap in a better proxy or, better yet, refer directly to what is really of interest to the criminal law. This is what the RCR does, and in doing so, it helps us address the problem of offences that appear to have overly broad conduct elements.
(b) Obviating artifice
The RCR also improves the law in another way – it obviates the need for some of the artifice that has crept into doctrine. Consider the famous case of Airedale NHS Trust v Bland.Footnote 57 With the consent of his family, doctors wanted to withdraw treatment, including food and drink, from a patient who had no hope of recovering from a persistently vegetative state. They sought the court's reassurance that they would not thereby be committing homicide. As a precursor to examining what was expected of the doctors, the House of Lords had to reach the somewhat strained conclusion that physically removing the nasogastric tube through which the patient was being fed was an omission rather than an act.Footnote 58 This was because, as the criminal law's conduct requirement for homicide is presently understood, our expectations of an agent are only relevant when her conduct is omissive. Once the removal of the feeding tube was dubbed an omission, the court could explain that the doctors were not expected (ie under no duty) to ‘act’ by ‘continuing to provide life support’. But if we adopted the RCR instead, our contextually salient expectations of an agent would always be relevant to determine whether she satisfies the criminal law's conduct requirement.Footnote 59 We could ask straightaway whether there was any contextually salient expectation that the doctors not withdraw the feeding tube. Theoretically, that expectation could come from law, or failing that, societal morality and routines (although on the facts, there was clearly no consensus on the existence of a societal expectation to keep treating such a patient). As such, under the RCR, the analysis would boil down to whether there were relevant legal expectations to keep treating Bland, which is exactly what the House of Lords addressed by reference to medical law.Footnote 60 Had the RCR been in place, the analysis in Bland would have been significantly simplified, and would not be predicated on the artificial claim that physically removing a feeding tube was not an act, but an omission.
Another artifice that the RCR would obviate is the ‘continuing act’ doctrine. In Fagan v Commissioner of Metropolitan Police,Footnote 61 D inadvertently parked on a constable's foot, but then decided to savour the constable's discomfort for a while before eventually moving the car. He was charged with battery, but proving the contemporaneity of actus reus and mens rea seemed problematic. When D drove onto the constable's foot, he lacked the mens rea for the offence. When he formed the mens rea, he was not actively doing anything. The Divisional Court took the view that being stopped on the constable's foot was a continuing act. This is clearly a fiction, but it allowed the Divisional Court to convict the defendant of a battery. At the same time though, it created all sorts of uncertainties about what sorts of individual pieces of conduct could be rolled up into a continuing act, how precisely one should delineate the beginning and end of a continuing act, and why.Footnote 62 None of these were addressed in Fagan, and when the continuing act doctrine has been raised in subsequent cases,Footnote 63 courts have preferred to sidestep the doctrine rather than attempt to clarify its scope. The RCR would let us finally put this notoriously vague doctrine to bed – clearly there was a societal expectation that Fagan move off the constable's foot once he realised that he'd stopped on it, and so his failure to do so was anomalous conduct that could satisfy the conduct element of the actus reus of battery at the time that he had the mens rea for the offence.
4. RCR and philosophical puzzles
I have suggested that the RCR clarifies and simplifies existing criminal law while offering us the argumentative resources to safeguard against the over-expansion of the net of criminal liability for innocuous doings, and to reconsider incongruities in the rules of omissions-based criminal liability. On the criminalisation side, the RCR helps conscientious legislators identify plausible candidate conduct-tokens for criminalisation and focus their thinking on the desirableness of criminalising those. Additionally, the RCR helps us make progress on two puzzles of enduring interest to philosophers of criminal law.
The first is the Trolley Problem. Many of the facts in the various statements of this problem (how many people are on each track, what their histories are, and so on) are not relevant to the issue of whether the agent's choice of conduct is eligible for criminal blame. They relate instead to her rationale-based explanations for her choice of conduct. But some responses to the Trolley Problem assert that only by pulling the lever (acting) does an agent author, and so become responsible for, the consequences.Footnote 64 Adopting the RCR entails rejecting that claim. All that matters for the RCR is whether the agent's chosen course of conduct belied a contextually salient expectation. That depends on the profile of contextually salient expectations in the agent's society, and different liberal societies may well have different such profiles. The relevant expectations consist of only imperative liberty-limiting directives, and exclude liberty-expanding permissions. A single fact can potentially generate either type of guidance. For instance, imagine that D is related to one potential victim in the Trolley Problem. A society might have a norm permitting D to favour her relatives over strangers. If so, then D's relationship with a potential victim increases the options available to her. We should therefore exclude it when deciding whether her conduct is eligible for criminal liability, though it would be relevant should D need to claim a rationale-based defence in respect of her chosen course of conduct. But if the norm is that D must favour her relatives over strangers, then D's relationship with the potential victim limits D's liberty, and so it is relevant to determining the eligibility of her conduct for criminal liability.
A given society might have various contextually salient expectations applicable to agents in moral dilemmas like the Trolley Problem. There may be
(a) an expectation that when faced with a moral dilemma, the agent will not intervene;
(b) no expectation specifically as to responses to moral dilemmas, but (only) an expectation that an agent will not intervene to cause death or injury; or
(c) no expectation specifically as to responses to moral dilemmas, but (only) an expectation that an agent will intervene to prevent or reduce the risk of death or injury; or
(d) no expectation specifically as to responses to moral dilemmas, but an expectation that an agent will not intervene to cause death or injury, and an expectation that she will intervene to prevent or reduce the risk of death or injury.
Each of these expectations may (and often will) be tempered by overriding liberty-expanding permissions, which would only be relevant if and when the agent provides a rationale-based explanation for her chosen conduct. Whenever the agent does not belie any contextually salient expectation applicable to her, her conduct will not be eligible for criminal blame. But in case (d) above, howsoever the agent (voluntarily) conducts herself, her conduct will be eligible for criminal blame. That, however, is a far cry from her being liable to conviction. Given prosecutorial discretion, she might never even be charged. If she is, a rationale-based defence would likely be available to her.
The second puzzle with which the RCR helps is the appropriateness of Bad Samaritan laws. The fact that different liberal states can have different sets of contextually salient expectations explains why it is perfectly coherent for some liberal states to have Bad Samaritan criminal laws, while others vehemently oppose them. But even people in liberal states that oppose Bad Samaritan laws should ask themselves, ‘Do we really not expect others to save us from serious harm when it would be easy and perfectly convenient for them to do so?’ If the answer is, ‘Actually, we would expect them to help’, then unless we can be certain that there is a relevant intrinsic difference in the normative significance of acts and omissions (which I doubt), these states have reason to reconsider their stance on Bad Samaritan laws. If there is no expectation of help, these states can remain secure in their current legal position. Such societies probably have bigger problems.Footnote 65
