Civilising loss of control? The role of criminal justice gatekeepers

Abstract

Since its inception in England and Wales, the partial defence to murder of loss of control has generated a steady stream of appeals. Individually, those appeals have illuminated key aspects of the plea's operation. This paper, though, is the first to explore that operation via a systematic analysis of every loss of control appeal to date (110 cases). Using that data, the paper frames more effectively, and thus improves understanding of, a neglected phenomenon in the plea: specifically, the decision-making roles of criminal justice ‘gatekeepers’ – principally trial judges, juries and prosecutors – in governing access to loss of control. In doing so, the paper assesses how far these gatekeepers interpret the plea's requirements in a ‘civilising’ way – one which prioritises meritorious loss of control claims above those which are unmeritorious. It contends that each gatekeeper struggles to regulate loss of control in such a way. Ultimately, this diminishes the symbolic value these reforms may have had and frustrates any civilising potential of homicide law reform.

Introduction

Many jurisdictions partially exculpate provoked killings. This has long been the case in England and Wales, where convictions for such killings may be downgraded from ‘murder’ to ‘(voluntary) manslaughter’.¹ That adjustment occurs despite the defendant (D) killing the victim (V) with the requisite mens rea for murder – an intention to kill or cause grievous bodily harm.² This mitigation operates via a successful plea of loss of control – a partial defence to murder contained in the Coroners and Justice Act 2009 (the CJA 2009). Loss of control has three requirements:

1. (1) a loss of self-control;

2. (2) caused by a qualifying trigger; and

3. (3) a person of D's sex and age, with a normal degree of tolerance and self-restraint and in D's circumstances, might have reacted in the same or a similar way to D.³

These requirements have some important features. In respect of (1), there is no need that the loss of control be sudden.⁴ In relation to (2), there are two prescribed triggers which may be claimed separately or in combination: whether D's loss of self-control was attributable to (a) fear of serious violence from V against D or another person, or (b) a thing or things done or said (or both) which constituted circumstances of an extremely grave character and caused D to have a justifiable sense of being seriously wronged.⁵ Finally under (3), recourse to ‘the circumstances of D’ is a reference to all of D's circumstances, other than those whose only relevance to D's conduct is that they bear on D's general capacity for tolerance or self-restraint.⁶

Loss of control's requirements raise broader questions of legal policy. Such questions relate to the doctrinal: about the proper boundaries of English homicide law, including loss of control itself; and to the theoretical: about the principles which ought to ground loss of control. They also concern matters of process: notably, the procedural and evidential hurdles which individuals should be required to surmount in claiming, and ultimately succeeding with, the defence. Indeed, since 4 October 2010, when the plea replaced its predecessor, provocation,⁷ loss of control's jurisprudence has burgeoned, offering fresh perspectives on these questions.

This paper delves much deeper into that jurisprudence via a systematic analysis of every loss of control appeal to date (110 cases). It thereby provides new perspectives on the defence's operation. This approach is unique, providing increased insights into the domestic regulation of provoked killings and much more so than a focus on selected appeals, as is common in existing loss of control literature. Importantly, then, the project extends knowledge of this area of English homicide law, akin to empirical research already undertaken into diminished responsibility,⁸ the other main partial defence to murder. The stakes are high: many individuals claim loss of control in negotiating the murder-manslaughter boundary and lessons on how the plea works in practice inform that boundary. Beyond England and Wales, such lessons may also be relevant to how loss of control/provocation pleas function in other jurisdictions – a comparative dimension which scholars continue to explore in identifying common developments and reform ideas.⁹

In this paper, we concentrate on process, ie the procedural and evidential realities of arguing loss of control. Here, we use data from our project to frame more effectively, and thus improve understanding of, a neglected phenomenon in the defence: specifically, the decision-making roles of criminal-justice ‘gatekeepers’ – principally trial judges, juries and prosecutors – in governing access to loss of control. In doing so, we assess how far these gatekeepers interpret the plea's three requirements in a ‘civilising’ way. This civilising framework has emerged as a prominent paradigm in evaluating the boundaries of appropriate criminal exculpation. A key exponent of that framework is Horder and we draw on his research regarding the civilising character of defences and, in particular, that of loss of control. By ‘civilising’, Horder means the extent to which official responses to wrongdoing (including the defining of defences) contribute to a more civilised and cultured, as well as more tolerant and humane, society; to that end, shaping the ways in which, and the attitudes with which, individuals flourish in common.¹⁰ The relevance of a civilising approach to regulating loss of control is further emphasised by the Law Commission of England and Wales (the Law Commission). During the reform process, it repeatedly stated that ‘unmeritorious’ loss of control cases should not succeed (no doubt because, under provocation, unmeritorious cases did occasionally succeed),¹¹ something which clearly invokes a civilising perspective. Our analysis indicates a lack of consistency in loss of control gatekeeping which has implications for the civilising function of the defence.

To begin, Section 1 briefly sets out our methodology. Section 2 then defines ‘civilising’ in the context of Horder's research on defences, including loss of control. As part of this, we highlight various factors cited by Horder as encouraging the development of civilising values in defences – values which we interrogate across Sections 3–5 in gauging how each gatekeeper interprets loss of control. Section 3 considers the first type of gatekeeper: trial judges. It analyses how they have exercised their role in deciding whether or not to leave loss of control to the jury. That role coalesces around a ‘sufficient evidence’ test which, far from contributing to civilising loss of control, is applied haphazardly, irrespective of a case's merits or demerits. Next, Section 4 addresses the second type of gatekeeper: juries. In determining the extent to which juries have played a more civilising role than trial judges, we contend that they sometimes privilege unmeritorious loss of control claims, as influenced by attitudes to gender and violence, as well as potential or threatened sexual violation of a third party. Lastly, Section 5 scrutinises the third type of gatekeeper: prosecutors. It argues that any civilising aspect to their role is compromised by the acceptance of loss of control in unmeritorious cases of male-on-male violence, thereby exacerbating the gendered nature of the defence.

1. Methodology

To develop the analysis, we read every Court of Appeal Criminal Division (CACD) judgment on loss of control since the defence's commencement. This involved carrying out legislative and keyword searches on the two major, online legal databases: Westlaw and LexisNexis. We included appeals against conviction or sentence, resulting in a yield – and subsequent analysis – of 110 judgments. Thereafter, the team examined the dominant themes that emerged from the CACD judgments, one of which was gatekeeping.

Necessarily, there are some limitations to this study. A concentration on CACD judgments presents only a partial view of loss of control in practice, depending as it does on the acceptance of leave to appeal against conviction or sentence. Whilst most of the CACD judgments provide some detail about the facts of a case, such judgments contain reduced information on the arguments at trial. We also anticipate that individuals who successfully argued loss of control – either at trial or by having a guilty plea to manslaughter accepted – are unlikely to appeal, unless they do so against their sentence (of which we had a few cases). A more forensic analysis of how the defence works requires empirical observations of homicide trials in England and Wales, together with access to relevant Crown Prosecution Service (CPS) data. This would also allow analysis of the role of another key gatekeeper – the defence – which we acknowledge plays a vital role in loss of control cases; however, the data in the CACD judgments did not provide sufficient insights into this dynamic. Nevertheless, our study is the first to engage in a systematic and empirical analysis of all CACD cases concerning loss of control, therefore enhancing knowledge of the defence in action.

2. Civilising defences

We can make progress with assessing loss of control by viewing it through a paradigm that has become influential in evaluating the boundaries of appropriate criminal exculpation: namely, the ‘civilising’ character of defences, as popularised by Horder.¹² We have already noted that,¹³ doctrinally, this means framing defences so that they contribute to a more civilised society. Significantly, such a civilising perspective also underpins the Law Commission's desire to eliminate ‘unmeritorious’ cases from loss of control. This is doubtlessly why, in contemplating the role of the judge in filtering out unmeritorious cases, Horder comments that:

[i]t cannot be doubted that this [reform] is concerned not merely with cases where the evidence is factually weak, but also with cases … where it would, in a civilized society, be morally wrong even to risk the chance that a jury might show leniency to D.¹⁴

For Horder, the civilising agenda involves balancing eight factors – which he terms ‘strategic concerns’ (or ‘common goods’) – that set limits to, as well as provide scope for, the definition of defences. These factors will override spurious pleas based on individual convenience, idiosyncratic conceptions of the good, special pleading etc.¹⁵ Of these eight factors, the five most immediately relevant comprise:

1. (i) Maintaining a close match between how a defence – and the prohibitions to which it applies – ought morally to be regarded by those whom it governs, and by those who must interpret and develop it, and how it is regarded (‘Matching Perceptions’).

2. (ii) Preventing the development of defences perverting or distorting, amongst citizens and officials alike, a proper understanding of the importance (absolute and relative) of personal and proprietary interests, along with public or common goods (‘Understanding Interests’).

3. (iii) Ensuring that the availability of defences does not undermine the promotion of a culture of compliance and law-abidingness, in relation to the state's good faith attempts to promote the common good through regulation (‘Ensuring Compliance’).

4. (v) Discouraging the emergence of a ‘defence industry’ fostered by experts and professional advisers, in relation to particular kinds of claim, of a kind likely to result in the success of too many unmeritorious claims (‘Discouraging Claims’).

5. (vi) The need to encourage citizens to seek redress through political or bureaucratic processes rather than resorting to ‘self-help’, especially where the latter entails the use of force (‘Avoiding Self-help’).¹⁶

Our analysis reveals that, while these five factors provide valuable insights into how loss of control plays out in practice, the three interrelated concepts of ‘Matching Perceptions’, ‘Understanding Interests’ and ‘Ensuring Compliance’ are particularly helpful in highlighting incongruity between the aims behind reforming the defence and how it is interpreted at different stages of the criminal justice process. Ultimately, these interpretations diminish the symbolic value these reforms may have had, particularly in relation to gendered violence.¹⁷

For Horder, there are few limits on which of the five factors can be adduced in restricting – and duly civilising – defences.¹⁸ Inevitably, what counts as a civilising development in a defence may be contested:¹⁹ it will depend on the political task of establishing and maintaining a (morally) sound legal order – one that sustains relevant common goods.²⁰ The nature of that order will be in line with reasonable expectations,²¹ meaning, necessarily, a degree of normativity in considering whether and how a certain matter relates to a defence. To that end, defences have a progressive part to play in moulding the values with which citizens identify as part of living together as a community. In the case of loss of control, Horder states that one such value exists in the block on using the plea where D's qualifying trigger (the second requirement) stems from things said or done which constitute sexual infidelity.²² This is a symbolic feature that addresses a distinct matter: gender relations.²³ Men can no longer avail themselves of the defence when confronted by their (usually female) partner's infidelity. This is because under, say, ‘Matching Perceptions’, it is deemed unacceptable to react to such provocation by losing control and killing. Thus, we can use a civilising strategy to identify fatal violence that is deemed normatively worthy of mitigation.²⁴

If anything, Horder asserts that the CJA 2009 could have gone even further in its value symbolism. Why not disqualify police officers from making use of loss of control when the killing took place in the execution of their duty; or disqualify as provocation anything said or done by a child under the age of 10?²⁵ This would have addressed, respectively, two distinct matters: tolerant and restrained policing, and the fact that infants are more likely than others to be the victims of homicide.²⁶ Both of these matters invoke ‘Understanding Interests’. Perhaps even the very existence of loss of control as a partial defence to murder is uncivilised, given that the loss of self-control (the first requirement) no longer needs to be sudden. That stipulation ‘opens the door to mitigation in too many cases where D has had time to brood on the supposed provocation. [It] undermines the ideal that in a civilized society people temporize, and restrain their desires for retaliatory suffering through violence’.²⁷ On this view, loss of control should be abolished as it conflicts too much with ‘Ensuring Compliance’ and ‘Avoiding Self-help’. Nonetheless, our focus is the ongoing legal reality of the loss of control plea, as interpreted by three types of criminal justice gatekeeper: trial judges, juries and prosecutors. We therefore now turn, across Sections 3 to 5, to investigating how each gatekeeper applies loss of control's three requirements according to values which recall Horder's civilising agenda.

3. The trial judge as gatekeeper

In adversarial trial systems like that in England and Wales, the trial judge's role has traditionally been one of passivity. Indeed, Roberts suggests that this passivity represents a key structural feature of English adversary trials:²⁸ non-interventionism is the order of the day.²⁹ Such an ‘umpireal’³⁰ view of the English trial judge is tenable up to a point, although it does require qualification. Trial judges necessarily retain some interventionist powers, thereby exercising a degree of control over the trial process.³¹ Notably, they discharge a vital sifting function regarding information presented to the jury, ruling on the legal admissibility of evidence offered by parties to the proceedings. In this way, English trial judges exert ‘direct, peremptory control over the jury's responsibility for fact-finding in criminal litigation’.³²

It is against this backdrop that the trial judge plays a significant gatekeeping role in loss of control. Under the CJA 2009, section 54(5) stipulates that ‘sufficient evidence’ of the defence is required before a jury can assume that loss of control is satisfied (unless the prosecution proves beyond reasonable doubt that it is not); whilst section 54(6) clarifies that ‘sufficient’ means enough evidence ‘on which, in the opinion of the trial judge, a jury, properly directed, could reasonably conclude that the defence might apply’. As Thomas suggests, the sufficient evidence test ‘refocuse[s] the responsibility and duty of the trial judge to scrupulously evaluate the evidence in determining whether the defence ha[s] satisfied the evidential burden of proof’.³³ Consequently, the trial judge is the arbiter of what surmounts that evidential threshold, and this implies active judicial deliberation – and discretion – on what satisfies the plea. That deliberation should be greater than it was under provocation, especially given the open-textured nature of the triggers in loss of control's second requirement (notably the need for circumstances of an extremely grave character which caused D to have a justifiable sense of being seriously wronged). In contrast, provocation only required ‘evidence’, however tenuous,³⁴ for the defence to be put to the jury, meaning that the trial judge was obliged to leave provocation to the jury, even where no-one could conclude that a reasonable person would have reacted as D did. A similar position continues to apply to complete supervening defences, where the trial judge must only withdraw a defence from the jury if there is ‘no evidence whatsoever to support it’.³⁵

(a) From ‘evidence’ to ‘sufficient evidence’: trial-judge gatekeeping expanded

Under provocation, the limited nature of the trial judge's gatekeeping role meant that the plea was often left to the jury – even in undeserving cases. An infamous example is R v Doughty,³⁶ where D was initially convicted for murdering V, his newly born baby, who was crying persistently. In quashing D's conviction, the CACD noted that ‘there is no doubt, and it is not in dispute, that there was here evidence upon which the appellant was … “provoked” to lose his self-control’.³⁷ Such evidence was easy to establish in provocation, especially because the loss of self-control test was not subject to any legally prescribed triggers.³⁸ Anything could validly provoke D's killing of V. Nonetheless, as with loss of control, provocation contained a requirement regarding whether a reasonable person would have killed V in the circumstances.³⁹ However, assuming there existed any evidence of a loss of self-control, this requirement would then be left to the jury – and this is precisely what happened in Doughty.

Predictably, Doughty caused a furore and a key aim in reforming provocation was to revitalise the trial judge's gatekeeping role. The Law Commission noted that the requirement to leave the defence to the jury did ‘not serve the interests of justice because … [it] increase[d] the likelihood that an unmeritorious claim may succeed’.⁴⁰ That link between ‘justice’ and the perceived merits of a case proved highly influential in formulating the sufficient evidence test in loss of control. To that end, the Law Commission recommended that the trial judge be tasked with ‘filtering out purely speculative and wholly unmeritorious claims’.⁴¹ Its reports gave various instances of cases it considered ‘wholly unmeritorious’: D's claim ‘that he was provoked to lose his self-control by V's failure to cook his steak medium rare as ordered’;⁴² behaviour demonstrating religious or racial bigotry which was considered ‘offensive to the standards of civilised society’;⁴³ or cases ‘where the provocation alleged is simple separation or infidelity’.⁴⁴ It is easy to see how these instances align with civilising concerns regarding what kinds of lethal violence may, and may not, be mitigated.

The version of loss of control which appears in the CJA 2009 reflects such concerns – as illustrated by the removal of sexual infidelity from the scope of the defence,⁴⁵ along with the plea's inapplicability where D acted in a considered desire for revenge.⁴⁶ Norrie thereby observes how the sufficient evidence test means that the trial judge has the gatekeeping power to remove cases from the jury's consideration, rather than always requiring the jury to consider ‘the issue of moral and political acceptability in being provoked’.⁴⁷ As noted above, Horder highlights that test as a key example of the civilising process of law reform.⁴⁸ However, he notes that the sufficient evidence test ‘comes at a moral cost’ as it ‘devolves to the trial judge in an individual case the role of the guardian of civilized values’,⁴⁹ rather than, say, the legislature or common law.

Yet the idea that the sufficient evidence test promotes civilised values is itself contestable. Whilst that rule may seem suited to tackling unmeritorious cases, our analysis suggests it is not. Although a claim which lacks sufficient evidence of loss of control will technically be without ‘merit’, that may nevertheless exclude otherwise meritorious cases – ie where there was a morally convincing reason to kill, but insufficient evidence as to one or more of the plea's three requirements. In comparison, meritorious cases that struggle to prove these requirements may still satisfy the sufficient evidence test, at the whim of a generous trial judge. Perversely, even unmeritorious cases may surmount the sufficient evidence threshold where they clear all three loss of control requirements. These various challenges emerge because ‘merit’ is conceptually wider – and vaguer – than ‘sufficiency of evidence’, the former invoking a strongly normative (and not simply evidential) measure. At first glance, normativity should inform the sufficient evidence test, especially given the existence under the second requirement of the ‘fear’ and ‘anger’ triggers. These circumscribe what may legitimately provoke D in ways that require moral assessment – the former trigger stipulating ‘serious’ violence and the latter trigger requiring ‘extremely grave’ circumstances which caused a ‘justifiable sense’ of being ‘seriously’ wronged.⁵⁰ To this end, the CACD comments that the existence of the triggers will prevent some of the more ‘absurd trivia’ from falling within the defence.⁵¹ Of course, under its third requirement, the defence also involves an external appraisal of D's conduct, duly extending the degree of normative enquiry in the plea.

(b) ‘Sufficient evidence’ v ‘unmeritorious’: first-instance decisions

Nonetheless, our data suggests that trial judges have neglected ‘merit’ at the expense of prioritising ‘sufficiency of evidence’. Their gatekeeping role is thereby non-normative. Of course, that role still requires ‘common-sense judgment based on an analysis of the evidence’,⁵² this requiring ‘a much more rigorous evaluation’ than was expected under provocation.⁵³ Clearly, each of the three loss of control requirements is integral to that task. These requirements should be analysed ‘sequentially and separately’,⁵⁴ although the fact that they are assessed non-sequentially will not, of itself, secure an appeal.⁵⁵

In deciding whether there is sufficient evidence, trial judges are to consider the ‘quality and weight’ of the evidence.⁵⁶ In practice, though, that assessment is not always a good proxy for assessing the merits of a loss of control claim. An evidential concern with demonstrating a loss of self-control – the first requirement – has tended to dominate proceedings. It is here that a tension emerges between gauging sufficiency of evidence and barring unmeritorious cases. Whether D lost self-control is a subjective enquiry and not something which invokes ‘merit’: either there is evidence (of sufficient quality and weight) that D lost self-control, or there is not. That subjectivity enables the possibility of unmeritorious cases demonstrating a loss of self-control where D simply lost self-control on the facts.

Moreover, given the well-known difficulties in defining a loss of self-control,⁵⁷ unmeritorious cases can still surmount this hurdle, even where the facts do not necessarily suggest a loss of self-control. For instance, in some decisions the defence was left to the jury even though D had sought out and killed V at V's own home. An example is R v Bowyer,⁵⁸ where D admitted going to V's house to commit burglary – hardly a sympathetic case (and arguably contravening most of Horder's civilising factors). Conversely, the same facts might obtain in an arguably meritorious case – as in Attorney General's Reference (No 103 of 2015) ⁵⁹ where D went to V's flat to confront him about sexually abusing D's children, with D changing her clothes beforehand and taking with her a knife, a hammer and a wrench. Sometimes, there may be meritorious cases which fail to satisfy the loss of self-control requirement and go no further. Possible examples include R v Magson ⁶⁰ and R v Labinjo-Halcrow,⁶¹ both of which featured women who killed abusive partners. These debates highlight a disconnect between the presence of sufficient evidence as to a loss of self-control and the supposed merits of a case. They also show that it is not unusual for meritorious or unmeritorious cases to fail at this first requirement, which in the former instance could contravene ‘Matching Perceptions’, suggesting a misalignment between the perceptions of trial judges and the legislator.

In our sample, trial judges rejected the defence argument that loss of control should be left to the jury in 39 cases – and in 24 of those cases, the trial judge found that there was not sufficient evidence that D had lost self-control. These findings acquire yet more significance given the Law Commission's preference for retaining the judicial gatekeeping role alongside abolition of the loss of self-control requirement. That abolition was rejected by the government, which insisted on keeping the loss of self-control requirement, even using it to inform the name of the new plea.

In considering how trial judges assess the quality and weight of evidence, it is also relevant to consider gender stereotypes. In the context of admitting sexual history evidence in rape trials, research shows that a trial judge's consideration of what may be considered relevant to a case perpetuates problematic perceptions of female sexuality.⁶² The same worries troubled the law of provocation, particularly due to the (now abolished) ‘suddenness’ requirement. The point here is that feminist concerns regarding judicial interpretation of evidential provisions apply equally to trial judges’ determination of the sufficient evidence test in loss of control. Whilst feminist criticisms regarding provocation's response to abused women who kill was a key driver for reform, this is undermined by retention of the loss of self-control element and its interpretation by trial judges. Such criticisms are particularly applicable when notions of ‘good’ or ‘common’ sense are invoked (which is the approach that trial judges are to adopt when evaluating the evidence). As Hunter has argued, such notions tend to be gendered, incorporating ‘differences based on race, ethnicity, class and sexual orientation’.⁶³ Consequently, ‘it is not logic but cultural beliefs’⁶⁴ that tend to determine judicial evaluations of what is sufficient evidence.

Accordingly, under the sufficient evidence test, anxieties about the impact of gender on the filtering out of unmeritorious cases are pronounced in cases of domestic homicide. In our data, there were 16 cases where the trial judge left loss of control to the jury where the killing took place in the context of an intimate relationship between D and V – most of which contravene ‘Matching Perceptions’, ‘Understanding Interests’ and ‘Ensuring Compliance’ at the very least.⁶⁵ This contravention occurs because, in these cases, the defence fails to address the gender bias which the reforms aimed to tackle: we see the continued mitigation of domestic abuse and male violence against women, along with the related failure to respond adequately to the experiences of abused women. In 11 of these cases, a man killed his female partner or ex-partner. In several of these cases, it is hard to discern a loss of self-control or a qualifying trigger (the first and second requirements). In R v Morgan,⁶⁶ V worked as an escort and D, described as a wealthy businessman, was one of her clients. D claimed that V financially exploited and blackmailed him. The trial judge left loss of control to the jury, despite significant evidence of premeditation (preparing a list of items to use in killing V and disposing of her body, gathering those items over several days), with D giving evidence that he tightened a ligature around V's neck twice before doing so a third time fatally, cleaning up the crime scene immediately afterwards and hiding V's body. In these circumstances, it is difficult to imagine when D could have lost self-control. Furthermore, in R v Wilcocks,⁶⁷ D strangled V, his partner, during an argument, releasing his grip when she became unconscious. An argument erupted after V regained consciousness and D strangled her a second time, this time fatally. Again, it seems difficult to see how D had lost his self-control when he killed V, yet loss of control was left to the jury on two occasions (since D's first conviction was quashed due to a misdirection and he was convicted of murder again on retrial). Additionally, in R v Brehmer,⁶⁸ D was having an extramarital affair with V who was threatening to tell D's wife. D strangled V, originally claiming that V had attacked him with a knife, but later admitting that the injuries were self-inflicted. In this case, whilst the loss of self-control might be inferred from the violence of the killing,⁶⁹ it is difficult to identify a qualifying trigger. R v Mehmedov presents a similar picture. Here, V had moved to a refuge and had been granted a non-molestation order against D who repeatedly stabbed her the day after a family court's decision about their infant son.⁷⁰ The wider context in this (and other) cases was that of relationship breakdown and/or alleged adultery,⁷¹ and while (on the available facts) these were not used as triggers, they imply that the killing of women by men is still considered potentially worthy of mitigation.

That the gendered dynamics of intimate or purely sexual relationships still seemingly influence judicial interpretation of the sufficient evidence test casts doubt on the civilising progress of loss of control. There is a strong sense that the judicial gatekeeping role has not been successful in discriminating between meritorious and unmeritorious cases. Under ‘Discouraging Claims’, this resonates with Horder's view that a particular civilising function of homicide law is a need to ward against a ‘defence industry’ whereby particular kinds of claim are likely to be unmeritorious.

(c) ‘Sufficient evidence’ v ‘unmeritorious’: appealing first-instance decisions

The Law Commission envisioned that restoring the trial judge's power to filter cases, ‘coupled with the supervision of the appellate courts’, would ‘enable the law to set boundaries in a reasoned, sensitive and nuanced way’ – thereby chiming with ‘Matching Perceptions’, ‘Understanding Interests’ and ‘Discouraging Claims’.⁷² This is because the judge's filtering role involves a normative judgement regarding the interpretation and application of the defence, which can be seen as aiming to deter unworthy cases. However, as with trial judges, our data suggests that the CACD's ability to correct poor normative judgements made at trial – and thus contribute to civilising loss of control – has not been realised.

Indeed, the CACD has had many opportunities to review the operation of loss of control. Those opportunities have arisen, notwithstanding the tendency of many trial judges to leave the plea to the jury, with this happening in 52 of our cases. This proportion is considerable given the sufficient evidence threshold and the fact that the people who did have their loss of control case put to the jury are perhaps less likely to appeal than those where the trial judge did not leave the defence to the jury.

Either way, as the sufficient evidence test involves a judgement call, this involves close control by the CACD of trial judge decisions, since the decision is either right or wrong, with limited margin of appreciation left to trial judges.⁷³ However, in practice, the CACD rarely quashes convictions where the trial judge withdraws loss of control from the jury, and the CACD has emphasised that it will generally trust the decision of the trial judge given their unique position of having heard all the evidence. Indeed, the CACD has only upheld an appeal against conviction on the basis that loss of control was not left to the jury in two cases (out of 39 where the plea was not left to the jury) – those of R v Clinton and R v Turner.⁷⁴ Both of these cases involved issues of sexual infidelity. In other cases, whilst some disagreement with the trial judge is apparent, that disagreement relates to the loss of control requirement upon which the claim fails, with the CACD otherwise concurring with the trial judge's overall decision not to leave the defence to the jury.

On that point, the preoccupation of the CACD with the first requirement, that of a ‘loss of self-control’, reveals some tension between the CACD and trial judges as to what constitutes sufficient evidence that D lost self-control. Here, we can see some clash with ‘Matching Perceptions’. For example, cases such as R v Dawson ⁷⁵ and R v Goodwin highlight differences between the trial judge and CACD in the extent to which forensic evidence of ‘frenzied’ violence may be indicative of a loss of self-control. In both these cases, whilst the CACD did not overrule either decision, it clashed with the trial judge's assessment that there was sufficient evidence that D lost self-control solely based on the gratuitous violence inflicted against V. Moreover, in Dawson the CACD criticised the trial judge for inferring that D had lost self-control upon finding sufficient evidence of a qualifying trigger – thereby conflating the first two loss of control requirements. Significantly, in both Goodwin and Dawson, the CACD stated that D does not have to raise loss of control for the trial judge to leave the plea to the jury, so long as there is sufficient evidence from which this can be adduced.⁷⁶ However, the CACD further acknowledged that such a failure to raise the defence may support the conclusion that it is not in issue, and therefore should not be left to the jury. In Goodwin, D argued self-defence and gave evidence that he remembered hitting V with a hammer three times, but claimed he had no recollection after that. The trial judge considered that D had not adduced any evidence suggesting that he had lost self-control but accepted that the loss of self-control could be inferred from the forensic evidence that the victim had received at least 18 blows to the head. In this case, the CACD disagreed with the trial judge's analysis of the forensic evidence and commented that the failure by D to assert loss of control was ‘a very powerful point against the issue arising in such a case’.⁷⁷

Yet, disproportionately weighting the loss of self-control element has meant that the CACD has focused less on identifying and filtering out unmeritorious cases. As with trial judges, the CACD first considers the quality and weight of the evidence of D's loss of self-control and then makes a judgement call as to whether a loss of self-control is technically present or not. That reduces any opportunity to fulfil a civilising role and define which cases should be left to the jury based on merit.

However, whilst the CACD sometimes decides that trial judges are too ready to find sufficient evidence of a loss of self-control, it adopts a less strict approach to finding sufficient evidence of a qualifying trigger.⁷⁸ For instance, the CACD has been inclined to find evidence of the anger trigger in cases where trial judges initially precluded this on the grounds of D's acting in response to alleged infidelity, as in Clinton. Moreover, in R v Dawes whilst the CACD agreed with the trial judge's decision to withhold the defence from the jury (citing a lack of evidence of a loss of self-control), it went on to clarify that inciting violence does not necessarily preclude a finding of either the fear or anger triggers, ‘unless [D's] actions were intended to provide him with the excuse or opportunity to use violence’,⁷⁹ something which arguably contravenes most of Horder's civilising factors. This was also the case in Goodwin, where the trial judge presumed that, if the jury rejected D's plea of self-defence, this was indicative that there was insufficient evidence that D was in fear of serious violence from V, thus precluding the fear trigger. However, as the CACD went on to emphasise, self-defence is not only rejected because the jury does not believe that D was responding in fear, but also because it might find D's response disproportionate, as is often the case where D seeks primarily to rely on self-defence (but where the response may be considered excessively violent).⁸⁰ Bowyer therefore appears something of an outlier. Here, the CACD strongly criticised the trial judge's decision to leave loss of control to the jury in circumstances where D confessed to being a burglar who deliberately targeted V's house and then killed V when V allegedly reacted violently to D's presence in his home. The CACD said that:

[e]ven taking the appellant's evidence at face value … it is absurd to suggest that the entirely understandable response of the deceased to finding a burglar in his home provided the appellant with the remotest beginnings of a basis for suggesting that he had any justifiable sense of being wronged, let alone seriously wronged. On that basis alone, one essential ingredient of this defence was entirely absent.⁸¹

However, in finding sufficient evidence of a qualifying trigger in some unmeritorious cases, the CACD is sending a message that violence might have been mitigated in those cases, if only D had lost self-control. That appears to undermine the legislature's attempts to demarcate more clearly the sorts of circumstances where violence is not worthy of mitigation, suggesting that the CACD is muddying the civilising boundaries of loss of control across all of Horder's five factors.

4. The jury as gatekeeper

The jury represents a key feature of common law criminal trials, including in England and Wales. Its gatekeeping role primarily relates to fact-finding, although as Roberts notes, ‘the interweaving of fact and law in a multitude of statutory and judge-made legal tests entrusts modern juries with the power to pass moral judgment on the conduct of the accused, as part of the process of deciding “what happened” in relation to matters in dispute’.⁸² Thus, where determination of criminal liability rests on legal tests containing normative standards, juries must decide whether D's conduct ought to be criminally censured, something which resonates with ‘Matching Perceptions’ (in fostering a close connection between those governed by the law and those applying it), as well as promoting ‘Understanding Interests’ (in allowing a portion of the community to mollify judicial understandings of when violence is worthy of mitigation). This type of normative enquiry frequently arises in English criminal law whenever juries have to make moral evaluations of D's behaviour. Obvious instances include the test for ‘dishonesty’ in theft and fraud, this requiring assessment of whether D was dishonest according to the standards of ‘ordinary decent people’;⁸³ and the meaning of ‘gross’ in gross negligence manslaughter, this demanding consideration of whether D's negligence was ‘so bad’ as to warrant criminal sanction.⁸⁴ As Redmayne notes, these examples ‘seem to involve community standards and a jury – a cross-section of the community – seems well placed to answer them’.⁸⁵

This moral dimension to the jury's gatekeeping role is additionally present in loss of control. The plea is imbued with normativity in its second and third requirements: that D's loss of self-control be triggered by fear or anger (or both), and that a person of normal tolerance and self-restraint might have reacted in the same or a similar way to D. There is, then, room for juries to play a civilising function in regulating access to loss of control, even if that function exists on a small scale – one whereby juries represent pockets of the public, as a moral community, in governing the practice of criminal law.⁸⁶

However, as already highlighted,⁸⁷ trial judges withdrew loss of control from the jury in 39 of the 110 cases appealed to the CACD. The data suggests that these withdrawals overwhelmingly arose because of insufficient evidence of loss of control, something which usurps the jury's normative gatekeeping function,⁸⁸ albeit legitimately, as implied in R v Jewell.⁸⁹ Nonetheless, that legitimacy appears to rest on the assumption that the sufficient evidence test is suited to extricating unmeritorious claims, which we have argued it is not.⁹⁰ In any event, and as similarly noted above,⁹¹ our data also shows that trial judges did leave loss of control to the jury in 52 cases appealed to the CACD. This reinforces the idea that the judiciary is generally reluctant to encroach upon what is regarded as the jury's area of decision-making, a practice which Laird notes in respect of diminished responsibility (the other main partial defence to murder).⁹² Moreover, that practice is one which legal practitioners support. In Fitz-Gibbon's interviews with criminal justice professionals, such figures felt that ‘juries inject an important community value judgement into the evaluation and categorization of differing culpabilities in homicide cases, a judgement that … should not be overridden by legislation’.⁹³

Yet how effective are juries at performing their normative gatekeeping role when asked to do so? Morgan argues that the high threshold for leaving loss of control to the jury, coupled with the standard of proof for the prosecution, would ensure that ‘if the defence is left to the jury, it will succeed’.⁹⁴ This perceived tendency for juries to accept the plea is also voiced in Fitz-Gibbon's research. Here, interviewed stakeholders argued that, when offered, juries will accept a partial defence as a ‘middle ground’ between acquittal and conviction for murder; this being due to a reluctance to label perpetrators as murderers while wanting to give prosecutors something.⁹⁵ Nevertheless, this tendency is not borne out by our data, with juries convicting D of murder in 40 cases out of 52 where loss of control was left by the trial judge. In contrast, it was only in 10 out of 52 where juries accepted loss of control where it was left to them by the trial judge.⁹⁶

Importantly, though, these 50 cases do illuminate how juries perform normative gatekeeping in relation to two key themes: gender and violence, along with sexual violation prevention. Moreover, those two themes intersect with the focus on ‘sufficient evidence’ and ‘unmeritorious’ cases discussed in Section 3. There, we discussed how the sufficient evidence test filters through to the jury some meritorious cases – even those where sufficiency of evidence is dubious; and that, conversely, the test also filters through to the jury some unmeritorious cases – including again where sufficiency of evidence is dubious. Problematically, the approaches of juries to gender and violence, together with sexual violation prevention, also often fail to track the meritorious-unmeritorious divide, duly questioning how far juries aid in the civilising of loss of control, particularly in relation to ‘Matching Perceptions’ and ‘Understanding Interests’. This brings into focus how the criminal law should respond to violence inflicted by men.

(a) Gender and violence

Interestingly, our data appears to indicate that juries are generally not susceptible to interpreting loss of control to excuse abusive men who kill their partners. In numerous cases featuring intimate or purely sexual relationships – frequently a man violently killing a woman – juries still convicted D of murder. Accordingly, the representative nature of juries as comprising 12 random people would seem to be doing its job: jurors can see past traditional opinions of masculinity in confronting partners or lovers, usually in the context of sexual matters. At first glance, then, we might say that juries do civilise loss of control along the lines of ‘Matching Perceptions’ and ‘Understanding Interests’, in that their views at times cohere with the reform aim to protect women from male violence. Also, the jury sometimes corrects the brute nature of the sufficient evidence test, with its imperviousness to merits, and the associated willingness of trial judges to leave these cases to the jury. As cited above,⁹⁷ such cases comprise Morgan, Wilcocks and Mehmedov, along with another: R v Davies-Jones.⁹⁸ They also include R v Taylor ⁹⁹ and R v Mason ¹⁰⁰ where juries returned murder convictions for killings committed in, respectively, female and male same-sex relationships.

However, caution is required. Contrastingly, the cases of R v Gale ¹⁰¹ and R v Mokadeh ¹⁰² suggest that juries still hold gendered attitudes which lead to acceptance of loss of control where men use violence against other men. In Gale, D successfully relied on loss of control after meeting V to discuss a debt that D owed him. V walked away before turning back towards D with an axe in his hand – at which point D armed himself with a wrench and hit V around the head with it. Having killed V, D then hid V's body and lied to friends, family and the police to conceal his crime. In Mokadeh, D stabbed V outside a house party. This took place in the wider context of violent behaviour on an estate between groups of men, along with disdain for those who ‘snitched’ to the police. In both cases, D sought to rely on self-defence and each time the jury rejected this – though the juries accepted loss of control. Both claims are arguably unmeritorious, raising questions about whether loss of control portrays male-on-male violence as culturally (if partially) tolerable in situations of excessive self-defence, and the civilising implications of this. Relying on the normative gatekeeping role of juries is further complicated by the degree to which they are reluctant to label sympathetic individuals as murderers. In this way, performance of gender roles is critical to how the jury perceives D and V: clearly, gender influences the jury's conceptualisation and understanding of appropriate loss of control in a way that contravenes reform aims regarding male violence. An extensive body of feminist literature has revealed not only the currency which gendered tropes retain in convincing juries, but also the pivotal role of trial judges in (re)framing these narratives.¹⁰³ Moreover, studies indicate that the trial judge's summing up often derives from and perpetuates identities of masculinity/femininity,¹⁰⁴ thereby influencing jury verdict construction – something which demonstrates the need for judicial training on conscious and unconscious bias.¹⁰⁵

Adherence to norms about gender and violence may also explain unmeritorious verdicts in Gale and Mokadeh, as well as R v Lodge. ¹⁰⁶ Here, despite inflicting essentially unprovoked violence on V, another man, by stamping repeatedly on his head, D is portrayed as broadly sympathetic. Meanwhile, his girlfriend who accompanied him, is presented as morally culpable for his violence, deriving from, and perpetuating, the trope of women as a ‘muse’ – in this case, the catalyst for male violence.¹⁰⁷ In the judgment, the CACD noted that ‘the judge accepted that [the girlfriend] bore significant moral responsibility for these events. If it had not been for her the applicant would not have been involved in this violence.’¹⁰⁸ A decision raising similar issues is R v Daicov,¹⁰⁹ where D was in the process of engaging the services of a sex worker in an alleyway. Another man, V, arrived and confronted D who then launched a sustained attack on V, including kicking V when V was on the ground. It appears that, despite reservations from the trial judge regarding the truthfulness of D's account, the jury was convinced by D's argument that V and the sex worker were acting in concert to steal D's wallet.¹¹⁰

What is significant about the male-on-male violence perpetrated in Lodge and Daicov is that, whilst the trial judges left loss of control to the jury (despite the apparent absence of the second requirement – ie a qualifying trigger – in either case), each trial judge nonetheless remarked on the ‘low degree’ or ‘low measure’ of provocation present.¹¹¹ This signifies that the judges felt compelled to leave these unmeritorious cases to the jury because, despite it contravening the judges’ own normative standards, they found there to be (just) sufficient evidence of loss of control. It may also be that they presumed that the jury's own normative gatekeeping function would result in a murder conviction – even though, ultimately, these cases show that juries will not always condemn what might be described as uncivilised and morally impermissible physical aggression (this factor highlighting issues in relation to most, if not all, of Horder's civilising factors).

Problematic attitudes towards gender do occasionally still arise outside male-on-male violence. In Brehmer, the fact that the jury accepted D's loss of control plea (D having strangled V, his lover of 10 years, in response to her threatening to tell his wife of their relationship) shows that juries also continue to mitigate male-on-female violence, potentially viewing V as further removed from normative gender roles than D. Contrastingly, juries do not mitigate female-on-male violence, thereby further challenging how effectively juries play a civilising role in regulating access to loss of control. In our study, there were three such cases, R v Hughes,¹¹² R v Cox ¹¹³ and R v Jackson,¹¹⁴ all featuring abused female partners who killed their male abuser. Each contained meritorious features, yet the women were ultimately convicted of murder. In Hughes, D had endured a history of mental and physical abuse by V, along with derogatory remarks made by V just before the killing.¹¹⁵ Similarly in Jackson, D described her relationship with V as marred with controlling or coercive behaviour, culminating in V taunting her as she went to commit suicide, resulting in her turning the knife on V, stabbing him four times. Meanwhile in Cox, D's relationship with V had been ‘difficult, stormy and complex’, culminating in V strangling D, whereupon D stabbed V.¹¹⁶ The provocation revealed in these sets of facts surely indicates the presence of either the anger and fear triggers (or both) – although research elsewhere suggests that loss of control manslaughter is generally not forthcoming in such cases.¹¹⁷

(b) Sexual violation prevention

In our study, juries were additionally willing to accept loss of control pleas in situations where D killed V in response to potential or threatened sexual violation of a third party. This chimes with the Law Commission's example of an Asian woman who returns home to find two white men attempting to rape her 15-year-old daughter. After the men shout racist abuse at her and run away, she gets a knife, chases after them and stabs one of them in the back several times, killing him.¹¹⁸ That example was one on which the Law Commission surveyed public opinion on sentencing, with 70% favouring a custodial sentence of no more than five years and over 40% believing a non-custodial sentence was appropriate.¹¹⁹ As Horder remarks:

the fact that there was such strong public (not merely expert or interest group) support for leniency in this kind of case was subsequently openly acknowledged by the Government to have been crucial to its decision to retain a provocation defence with a basis going beyond an over-reaction to a threat of serious violence.¹²⁰

The Law Commission reinforced the need for a provocation plea in this sort of scenario by stating that if such a plea was restricted to cases in which D killed V out of a fear of serious violence, then it would be unavailable if, when chasing after the men, the woman no longer fears at that moment that they will do serious violence to her or her daughter.¹²¹

Our data indicates that jurors are indeed inclined to accept loss of control where potential or threatened sexual violation arises – something which contradicts ‘Ensuring Compliance’ and ‘Avoiding Self-help’ on the basis that D ought not lose control in a situation where other legal and non-violent avenues are readily available. This is evident from R v S where the jury convicted D, a young man highly protective of his older sister, of the manslaughter of V, his sister's boyfriend, after D witnessed V having what D erroneously thought was non-consensual sex with D's sister.¹²² However, it can be argued that there was no obvious second or third requirement (ie a qualifying trigger and need for a person of normal tolerance and self-restraint to have reacted in the same or a similar way) in this case. The lack of either a clear qualifying trigger, or the reaction being that of a person of normal tolerance and self-restraint, is also present in R v Malik where D killed V, his daughter's former mother-in-law, following rape threats made by V's son (the former husband of D's daughter) against another of D's daughters. Interestingly, S and Malik both indicate that violence is worthy of mitigation if used in order to prevent sexual violation, notwithstanding any considered desire for revenge that may be present on the facts – something which should bar a loss of control claim.¹²³ Certainly, in S, this level of revenge is arguably present, given that D hunted V down and, on discovering him, stabbed him numerous times, and admitted to being angry and protective; as it is in Malik, where D embarked upon a sustained episode of serious violence against his daughter's mother- and father-in-law, having fetched an axe and driven to their home, killing the former. Similarly, the jury accepted loss of control in Attorney General's Reference No 103 of 2015, where, following a carefully planned attack, D stabbed V, a man who had sexually abused her children (this was the only case in our sample where a woman successfully claimed loss of control). In all these cases, the successful use of loss of control in situations where not only are the second and third requirements ostensibly lacking, but also there is perhaps (a considered desire for) revenge, shows that there is difficulty in discerning a civilising approach from juries to loss of control.

5. The prosecutor as gatekeeper

Prosecutors play a similarly important role as gatekeepers in loss of control. Nevertheless, as with discussions pertaining to the trial judge and the jury, it is hard to discern consistency in determining whether prosecutors would accept a loss of control plea or not. This raises an issue with ‘Matching Perceptions’ regarding the disconnect between the aims, interpretation and application of the defence. That finding also needs to be placed in context: there is a relative dearth of appeal cases in which a guilty plea was accepted. There were 27 cases in our sample in which D either entered or formally indicated a guilty plea to manslaughter. However, the CPS accepted a guilty plea in only seven of these cases (this sits in contrast with the overall high number of instances outside loss of control where cases proceed on the basis that a guilty plea has been entered and accepted¹²⁴). That is perhaps not surprising though, given that people who had a manslaughter plea accepted by the CPS are less likely to appeal. Nonetheless, the loss of control cases indicate some reluctance from the CPS to accept pleas, with a preference for the CPS taking its chances with a murder trial – something which is perhaps not surprising given the lack of clear rules delineating which cases are likely to be successful. For instance, in three of our cases where the CPS did accept a guilty plea, it only accepted the plea very late. In R v Caddick,¹²⁵ the CPS accepted the plea after two days of evidence; in R v West,¹²⁶ the plea to manslaughter was only accepted after a first trial for murder where the jury was unable to reach a verdict; similarly, in R v Rawson,¹²⁷ the guilty plea was only accepted after the first trial for murder aborted.

(a) Mitigating male violence?

However, there are concerns with respect to the merits – or lack thereof – of those cases in which a loss of control plea was accepted. The cases demonstrate a problematic leniency towards male violence, a theme which has arisen in our analysis of CACD judgments in Sections 3 and 4. It is fair to say that these sorts of case are not emblematic of loss of control, highlighting tensions with most of Horder's civilising factors, with sentencing judges commenting on the brevity of the loss of self-control, or the low degree of provocation leading to the killing. This suggests that if such cases had proceeded to trial, the trial judge may have had some concerns about leaving the defence to the jury, although of course that is not certain.

Most of the cases involving male-on-male violence where the CPS accepted a loss of control plea could be considered unmeritorious. For example, R v Ward ¹²⁸ involved a drunken fight between friends who had spent the day drinking together and had gone back to the flat of one of them to carry on drinking. Violence erupted following a disagreement and D responded to his brother being headbutted by V by attacking V with a pickaxe handle. In R v Duncan,¹²⁹ D killed V, his girlfriend's ex-partner. Having seen V in the city centre, D went to buy a knife before initiating a verbal altercation with V. When V physically attacked D, D stabbed V fatally. A final example is that of R v Bird,¹³⁰ in which D killed V, his long-term female partner. The killing took place in a context where the police had responded on several occasions to allegations of domestic violence when both D and V had been drunk, but no prosecutions had ever been brought. The trial judge accepted that there was evidence that D had lost his self-control (presumably because of the attack being described as ‘frenzied’), although it is difficult to discern a qualifying trigger. The CACD noted that ‘in the judge's view nothing in the history of the relationship constituted behaviour on the part of the deceased that could be relied upon in advancing the defence of loss of control’.¹³¹ Although it is not expressly stated, it seems that a possible qualifying trigger could have been fear of serious violence, with D stating that V had hit him and was the first to pick up the knife. However, D also admitted that he had managed to disarm V easily,¹³² therefore placing in doubt the veracity of his argument. Finally, we can question whether the case would satisfy the third requirement of loss of control. Taken together, acceptance of loss of control in cases like Ward, Duncan and Bird hardly represents a shift towards civilising the loss of control defence.

Only two cases in which a guilty plea was accepted indicate potentially meritorious circumstances – notably, those of Caddick and R v Adeniran.¹³³ In Caddick, V had sexually abused D as a child and had served a prison sentence for his crime. It is interesting to note, however, that the trial judge perceived there to be a low degree of provocation (based as it was on V's assertion that ‘I did prison for you’ and not the abuse per se) and passed sentence (of eight years and eight months, upheld on appeal) on the basis of both loss of control and diminished responsibility. In Adeniran, D armed himself and killed V in the process of V and others unlawfully barging into D's home in their drunken and armed pursuit of another man. While self-defence was unsuccessful – possibly on the basis that D, having once stabbed V in the flat, chased after him and stabbed him again – loss of control was accepted. In this case, a sentence of nine years for manslaughter was upheld.

Of those cases in which the CPS did not accept a plea, the jury nevertheless found the offender guilty of loss of control manslaughter in five cases, most of which have been discussed earlier: Malik, Brehmer, Mokadeh, Daicov, along with a case not yet discussed – Attorney General's Reference No 29 of 2014, in which D stabbed V, a love rival.¹³⁴ Again, the key theme with these cases is male-on-male violence, the juries’ decisions thus mitigating the infliction of fatal harm that might nevertheless be considered deeply morally problematic and therefore unmeritorious. This sits in tension with ‘Matching Perceptions’, ‘Understanding Interests’ and ‘Ensuring Compliance’, as there is a disconnect between the reform's aims and the plea's interpretation in respect of when male violence is worthy of mitigation or not. Indeed, nothing particularly distinguishes these cases with those in which the CPS accepted a plea of guilty to loss of control. As a result, it is hard to discern any norm-setting strategy behind the CPS's decision-making process, and difficult to see what civilising approach prosecutors take, if any, in identifying unmeritorious cases.

(b) Gendered nature of guilty pleas

It is worth noting that everyone who had their plea to manslaughter accepted by the CPS was male. This can be explained by the fact that men represent the vast majority of defendants in homicide cases. Moreover, female defendants kill in different circumstances and most of those in our sample claimed self-defence, with loss of control only featuring as an alternative.¹³⁵ Given that self-defence is a complete defence, it is likely that defence teams were not prepared to discuss a guilty plea to manslaughter, preferring to put their case for self-defence to the jury in the first instance. Nonetheless, research indicates that such pleas are unlikely to succeed,¹³⁶ ‘either because the force used was disproportionate or she is unable to prove that the threat was imminent’.¹³⁷ Hence, the sagaciousness of running self-defence over loss of control can be queried. Despite this, though, research also indicates that loss of control is generally not being successfully raised at trial level, and that the CPS is reluctant to accept a plea to loss of control in cases involving abused women,¹³⁸ which accords with our analysis.

Indeed, there appears to be an overall propensity for abused women to receive a murder conviction, arguably contravening ‘Matching Perceptions’ and ‘Understanding Interests’ as the defence fails to acknowledge the extreme circumstances in which women tend to use fatal force. This is evidenced by Edwards’ review of cases involving female defendants convicted for killing an intimate partner, between April 2011 and March 2016. Out of 40 cases, she discloses that 30 women were convicted of murder, three of diminished responsibility manslaughter and another seven with ‘other’ manslaughter, which includes both loss of control and ‘lack of intent manslaughter’.¹³⁹ Edwards notes, however, that the publicly available information did not suggest that in any of these seven ‘other’ cases were the defendants convicted of loss of control by virtue of the fear or anger triggers.

Conclusion

This paper offers key, original insights into loss of control, by providing the first systematic and empirical analysis of all CACD judgments. We have focused specifically upon the gate-keeping role of three key criminal justice actors: the trial judge; the jury; and the prosecution. In particular, we have explored the ‘sufficient evidence’ test and the extent to which this has been successful in filtering out unmeritorious cases. To aid this analysis, we have drawn upon Horder's civilising process of law reform and highlighted the moments at which the three gatekeepers potentially contravene the various civilising factors, or strategic concerns.

Throughout, we have argued that the extent to which the sufficient evidence test promotes Horder's civilising factors and effectively filters out unmeritorious cases is disputed by our data. We have seen that trial judges frequently detach merit from sufficiency of evidence. Relatedly, the requirement that D suffer a loss of self-control has skewed the focus away from a more normative evaluation of cases, with many failing at this first evidential hurdle; both at trial and appeal level. At the same time, we have observed that while the CACD overwhelmingly agrees with the decision of the trial judge, it has nevertheless adopted a more generous interpretation of the fear and anger triggers. This, of course, broadens the scope of what might be considered worthy of mitigation – which may be considered to militate against the civilising potential of loss of control. Indeed, we have seen potentially unmeritorious cases being left to the jury – those in which evidence of loss of control is, at best, scant, particularly in cases of male violence which raise questions of gender stereotypes. Conversely, we have also seen the exclusion of possibly meritorious cases, particularly those involving female defendants who have killed their abusers. All these elements undermine and frustrate any civilising potential of homicide law reform.

Overall, it is difficult to discern any consistency in approach amongst and between the gatekeepers. This is problematic given the important role that loss of control plays in demarcating the line between murder and manslaughter. Indeed, it may render uncertain any legal advice given to defendants and erode confidence in the criminal justice process. Undoubtedly, more empirical research into the operation of the defence is needed. However, disparities between gatekeepers highlight tensions between the aims underpinning reform of the defence and how it is interpreted and applied in practice. Such tensions are particularly apparent in the tendency of the defence to mitigate male violence and its failure to recognise and respond adequately to domestic abuse that women suffer, whether as a victim or an offender. These different interpretations diminish the symbolic value these reforms may have had, frustrating any civilising potential of homicide law reform.