comment

There Can Be Only Two (Verdicts): The Presumption of Innocence and Jury Verdicts in Criminal Trials

In a recent article, Chalmers, Leverick and Munro argued that Scotland’s third verdict, that of ‘not proven’, is probably compatible with the presumption of innocence. In this short reply, I reach the opposite conclusion, and argue that the presumption requires a system of binary verdicts.

Jackson Allen

In the latest instalment of their rightly lauded series of publications on Scottish juries, James Chalmers, Fiona Leverick and Vanessa Munro concluded that ‘it is unlikely that the not proven verdict would be found to contravene Article 6(2)’ of the European Convention on Human Rights (ECHR). Article 6(2) is, of course, the ECHR’s  presumption of innocence (PoI) provision. ‘Not proven’ (NP) is Scotland’s oft-criticised third verdict, which results in acquittal of the accused. Despite its functional equivalence to an acquittal, the NP verdict remains as an option in criminal trials, with its exact meaning not being entirely clear. On its face, NP’s compatibility may seem inconsequential, given that the authors go on to argue for its elimination, comprehensively demolishing the three most prominent arguments in its favour. Indeed, I make no claim to rebut the overall thrust of their argument; nor would I wish to – I agree with their main conclusions.

However, giving credence to NP’s compatibility with the PoI is something that needs to be combatted. This is not just because it is arguably a missed opportunity for the authors to put their main argument about abolishing NP more forcefully, but also because it appears to misunderstand the PoI’s requirements with respect to jury verdicts. As I argue below, the PoI is not indifferent to the existence of the third verdict in Scots law. On the contrary, the NP verdict should be seen as flatly incompatible with the PoI.

Before developing this argument in greater detail, however, a brief summary of Chalmers, Leverick, and Munro’s argument with respect to NP’s Article 6(2) compliance is in order.

The authors reach their conclusion about NP’s probable compatibility with the PoI on two principal grounds. First, the authors observe that the NP verdict, though functionally equivalent to an acquittal, is otherwise largely meaningless. Or, perhaps more precisely, the NP verdict has no settled meaning, either in law or as a matter of public opinion. As the authors point out, some view the verdict as partially stigmatising, others as no different from an acquittal, and others still as an acquittal but with a ‘nudge and a wink’, as if to chide the accused, saying ‘you got away with it this time’.

Second, and perhaps more importantly, the authors note that ECHR jurisprudence on Article 6(2) is couched in terms that prohibit state officials from casting aspersions on the validity or appropriateness of an acquittal. This is what the European Court of Human Rights (ECtHR) held, for example, in Allenet de Ribemont v France (1990) 20 EHRR 557. But, as the authors argue (at 869), the NP verdict does not cast doubt on the acquittal, it just is the acquittal. Can this be correct? In what follows, I aim to show that it cannot be. This will then lead to the identification of a broader point, which is that the PoI mandates a system of binary verdicts in criminal trials.

The reason for this conclusion, as will be explained below, is that the PoI recognises and protects a status quo for all citizens: that they are innocent of criminal wrongdoing, and therefore at liberty. To rebut this presumption, and therefore to change the status quo with respect to an individual’s guilt or innocence of a given offence, there is one, and only one, legal process in place: proof of guilt beyond a reasonable doubt. Absent this, even in a trial in which serious questions are raised about the accused’s (moral) innocence, or their character more generally, the PoI requires that the status quo ante is restored. This means that the accused is legally recognised as being in precisely the same position as they were before, namely, the position of being not guilty of the offence charged. As such, the PoI mandates a system of binary verdicts, one recognising guilt, and the other its absence.

To show that the availability of a third NP verdict is incompatible with the PoI under Article 6(2) and as a more general principle, I will advance two key claims. The first is that, in order to reach the conclusion that they do, Chalmers, Leverick and Munro rely on a distinction which is superficial at best. This is the distinction between a) the post-trial disparagement of the accused’s acquittal, and b) the potential stigma which might follow a verdict of NP. They distinguish between these two cases, arguing that ECHtR jurisprudence on Article 6(2) extends only to the former. Second, I will argue that the PoI requires that only two verdicts be available at the end of a criminal trial: one recognising guilt, and one recognising its absence.

I. Article 6(2) and Jury Verdicts

The first claim can be dealt with relatively quickly, and has already been described in the preceding paragraph. Citing Allenet and two other cases (links here and here) the authors conclude that the decisive issue is that Article 6(2) prohibits state officials from casting doubt on the accused’s innocence after they have been acquitted. They contrast this with a requirement not to do so via the verdict, as is potentially the case where the accused receives a verdict of not proven.

The authors appear to distinguish the latter situation, which would effectively find NP incompatible with the PoI under Article 6(2), on the basis of its contemporaneity with the trial and its conclusion. In fact, however, there is very little to suggest that the timing of the casting of doubt ought to be decisive. On the contrary, in O v Norway (unreported), the Strasbourg court emphasised the relative unimportance of exact timing (at paragraphs 33 and 39–41), holding Norway liable for a breach of Article 6(2) for acquitting an accused of sexual offences but nevertheless making him pay damages to his accuser. Though the authors do refer to this judgment in their article, they have perhaps underestimated the force of its underlying reasoning in undermining the case for NP being compatible with the PoI. Moreover, the unimportance of timing is also reinforced by Allenet, mentioned above, where a government minister and several senior police officers made statements doubting the accused’s innocence before the trial had concluded. This was held to breach the PoI under Article 6(2) because the statements amounted to a formal declaration that the defendant was guilty. As such, it is this more substantive issue which seems to be the focus of the Strasbourg court with respect to Article 6(2)’s application.

In any case, as a matter of principle, it would be odd to insist that timing could be so decisive with respect to the requirements of Article 6(2)’s prohibition on casting doubt on acquitted individuals’ innocence. A potential real-life example might be helpful here. Consider the case of a local official who comments on a criminal trial in one of the three following ways:

  1. The official waits until the accused has been found ‘not guilty’, then holds a press conference the next day telling the public, “Despite the verdict, we all know that he did it, and can’t be trusted”.
  2. As above, except the accused is acquitted and the verdict is one of ‘not proven’.
  3. As above in Option 2, except the official holds the press conference as the jury verdict of ‘not proven’ is read out in court, expressing his view just five seconds after the NP verdict is pronounced.

Option 1 is plainly a violation of the PoI under Article 6(2), as was held in Allenet, the facts of which are largely similar to the scenario in Option 1. What about Options 2 and 3? Whatever view one takes about Option 2, it seems clear that there is no meaningful difference between Options 1 and 3. Therefore, if we are prepared to find that Option 3 breaches the PoI, we should also find that Option 2 does. In my view, this is the case, and all three options are incompatible with Article 6(2) as currently understood. In other words, the timing of the expression makes no difference here.

Rather, what is important is that a) the accused has been acquitted, and b) a state official has given the public the impression that they are not really deserving of such acquittal. What then distinguishes the situation in Option 3 above from a situation where the acquittal and the official communication are simultaneous, because the former entails the latter? The answer is that there is no principled difference. The simultaneity of the acquittal and the official communication, which in the case of NP comes from implication, rather than an express statement, should not make a difference here.

Alternatively, another way of reading Chalmers, Leverick and Munro’s argument is that there is a substantive difference between the act of treating someone who has been acquitted as if they are (at least partially) guilty by doubting their acquittal, compared to simply refusing to fully acquit them at the end of the trial itself. If this is their contention, it is not entirely clear why we should think of these two cases as so dissimilar that Article 6(2) applies to one and not the other.

On the contrary, the latter situation of refusing to fully acquit someone whose guilt has not been proved beyond reasonable doubt seems to be a clear-cut violation of the letter of Article 6(2), given that this provision holds that everyone charged with an offence is presumed innocent ‘until proved guilty’. In any case, as I argue below, the best understanding of the PoI is that it requires binary verdicts. If the present understanding of Article 6(2) is incompatible with this understanding (and I don’t believe it is), then I suggest it is the ECtHR whose interpretation is out of step with the letter and spirit of the PoI.

II. The PoI and Binary Verdicts 

The above argument raises a more general and fundamental question: what does the PoI require with respect to the number and content of jury verdicts in criminal trials? I now want to make the case that the PoI requires a system of binary verdicts, with two options: one for guilt, and the other for its absence. In what follows, it should be noted that I refer not to the relatively thin conception of the PoI as offered in Article 6(2), but rather to the PoI as a general principle of the criminal law.[1] Therefore, to the extent that it might be argued that current ECHR case law diverges from what I go on to describe, my view is that it is the ECtHR’s interpretation of the presumption which ought to change, because in such conflicts it is the PoI qua general principle which must be respected. Nevertheless, there is little reason to think that the views sketched below could not be accommodated within existing ECHR jurisprudence, even if that is not already the case.

The starting point here is that most systems of criminal law, especially in the common law world, already have a system of binary verdicts of the type described above. English law is a canonical example of this. One can either be found guilty, or not guilty, and that is it. This is the case even where the insanity defence is successfully proved, with the verdict being not guilty by reason of insanity.[2] But it does not follow that this is the right way of setting up a system of verdicts. An alternative, explored by Federico Picinali in a recent essay, is to have a system of ‘many valued verdicts’ (MVV) where there are a multitude of different verdicts. These may refer to different levels of confidence in guilt, such as 50 per cent sure, 75 per cent sure, and so on, or to some other different set of discrete value judgments. In Picinali’s view, a MVV system is compatible in principle with any of the leading theories of punishment and may have advantages over the binary verdict system. The connection between this and the ‘not proven’ verdict I will return to in a moment. But a prior question needs to be answered first, about the compatibility of an MVV system with the PoI.

Picinali claims that it is compatible with the PoI. In footnote 14 of his essay, he argues that the PoI does not require binary verdicts, because any such requirement would be ‘question-begging’, requiring us to define ‘innocence’ in such a way that presuming it necessarily entails that one is presumed innocent of anything short of guilt. The implication of this argument is that the PoI does not already define innocence in such a way, hence we cannot rely on the presumption to justify a system of binary verdicts without begging the question. It is not obvious, however, why this is the case.

In fact, the claim that an MVV system is compatible with the PoI as a general principle appears much more fraught than the claim that the PoI requires binary verdicts. This claim about MVVs is not only fraught but also, in my view, mistaken. The simple reason for this is that, as Gerry Maher argued in his 1983 article, such an approach incorrectly assumes there is ‘logical space between proof of guilt and proof of innocence’. In fact, as Maher pointed out, there is no such space. In other words, the issue of guilt or innocence in an adversarial trial is already predicated on a ‘bright line’ type of rule. One is not partially guilty or partially innocent at the end of a trial, at least not as a matter of law.

This is not to say that the public may not draw their own opinions about the moral innocence of the accused, even after a not guilty or not proven verdict. They certainly may. But it does not follow that the law should facilitate this type of scalar thinking about guilt and innocence in the specific context of jury verdicts in criminal trials. On the contrary, there are good reasons, grounded in the logic of the PoI, why we should think in binary terms about guilt and innocence as a matter of law.

This gets to the heart of the issue under consideration here, which is what the PoI requires with respect to verdicts in criminal trials. As Maher noted, the effect of the PoI is to collapse the distinction between failure to prove guilt and proof of innocence. The PoI presumes that the accused is innocent, and so there is no need to positively prove it; that is the whole point. Rather, the prosecution must bear the burden of proof with respect to the issue of guilt. This can either be proved, or not. The verdicts of guilty and not guilty therefore correspond to these two outcomes: the accused can be guilty, or not. There is no intermediate outcome. Regardless of whether jurors hold a view that the accused might be guilty, or a little bit guilty, the criminal law draws a bright line with the reasonable doubt standard. Thus, the jury cannot simply refuse to decide one way or another – they must reach a verdict on the basis of whether or not this standard is met.

Crucially, the reason that the jury must reach a verdict is the PoI. It is the PoI that establishes the default outcome for a trial in any circumstances other than proof of guilt to the requisite standard: an acquittal. This is why, as Maher notes in his article cited above, Lord Sankey in the his famous judgment in Woolmington v DPP [1935] AC 462, holds that if the prosecution cannot prove guilt the accused is ‘entitled to an acquittal’ (emphasis added). The accused is entitled to this by virtue of the PoI. As such, if the prosecution fails to prove guilt, the accused is legally exonerated, and the verdict must reflect this. Their pre-trial innocence, in the sense of the pre-trial status quo ante, is thus reaffirmed, even if public perception has shifted.[3] As such, there is nothing question-begging about reading a requirement for binary verdicts into the PoI. The requirement is already there.

Conversely, the alternative MVV system of the type described by Picinali, artificially creates space between proof of innocence on the one hand, and failure to prove guilt on the other. In so doing, it contravenes the basic requirement of the PoI, which holds that the accused can be found guilty if, and only if, they are proved guilty of an offence beyond a reasonable doubt. This is simply the logic of the PoI and its rationale as a principle of the criminal law. One is innocent unless and up until convicted by a court, following a fair trial. In the same way that one cannot be partially convicted of a single offence, one should not be able to receive a verdict which sits somewhere between guilty and not guilty. As such, the MVV system is not compatible with the PoI.

III. Not Proven and the PoI 

If one accepts my argument so far, then the only step left to realise my original ambition is as promised to return to the connection between MVV and NP, and to do so in order to show that Scots criminal law, with its third verdict, is currently an MVV system. In fact, this is less straightforward than it might seem. As Chalmers, Leverick and Munro point out, the meaning of the NP verdict is by no means clear. This is, as they correctly observe, one of many reasons to disfavour the third verdict. In essence, the NP verdict is ambiguous: it is either exactly the same as a ‘not guilty’ verdict, or it is something else, some sort of ‘nudge and wink’ type of middle ground, as the authors put it.

As a matter of law, of course, NP functions as an acquittal exactly the same as ‘not guilty’ does. But the message it communicates is less clear. If it communicates precisely the same thing as ‘not guilty’, then it is redundant, but nevertheless PoI-compatible. This would mean that the Scottish system does not really use multi-value verdicts, but rather has a poorly constructed binary verdict system. Alternatively, if NP does communicate anything other than a full acquittal, then Scotland has an MVV system which is incompatible with the PoI, as shown above.

In my view, it is more likely than not that NP represents a true third verdict, communicating something midway between guilt and acquittal. This is at least partially supported by the mock juror views quoted in the authors’ MLR article, which suggest that jurors choose NP when they want to communicate that the accused has done something wrong, and maybe even committed the offence in question, but that the standard of proof has not quite been met. It also makes intuitive sense that NP would have some meaning other than that communicated by a not guilty verdict. Admittedly, this is partly undermined by evidence that the verdict’s survival is largely an historical accident, as Chalmers, Leverick and Munro have previously argued. However, the case remains that its modern meaning could still diverge from the original reasons for its introduction to Scots law. As such, it seems that the modern NP verdict is conveying something different from the other two verdicts in Scots criminal law, some intermediate position between guilt and acquittal.

IV. Conclusion

Returning now to the original question this short essay set out to address, some brief answers can be given. Scots criminal law currently has three verdicts, meaning that it most likely qualifies as an MVV system. Such systems contravene the PoI by unduly recognising logical space between failure to prove guilt and proof that the accused is innocent, when such space does not exist. In so doing, these systems undermine the PoI and are incompatible with it. As such, NP is not, in fact, compatible with the PoI, insofar as the PoI under Scots law reflects the PoI as a general principle of the criminal law. Moreover, the NP verdict most likely breaches the PoI under Article 6(2), as currently understood, for the reasons laid out above.

The upshot of this critique is that there is an opportunity to make an even more powerful repudiation of NP than that which Chalmers, Leverick and Munro have already offered: it runs counter to an axiomatic and fundamental principle of the criminal law. This simply adds force to an already overwhelming argument in favour of abolishing NP. In the current political climate, the Scottish Government and Scottish Parliamen appear to have an appetite for criminal justice reform. It is therefore hoped that those in Holyrood with the power to effect this change will heed the call, and cement Scots law’s status as a PoI-compatible system.

References

[1] Elsewhere, I have referred to this as the ‘thick’ PoI. For more on the thick/thin distinction, see J Allen, ‘Rethinking the Relationship between Reverse Burdens and the Presumption of Innocence’ (2021) 25(2) International Journal of Evidence and Proof 115-134.
[2] Detailed discussion of the special verdict of not guilty by reason of insanity (NGRI) is beyond the scope of this brief forum post. But in simplest terms, I view NGRI as being a verdict with reasons, rather than a separate third option of the type distinct from ‘guilty’ and ‘not guilty’ simpliciter. Thus, the special verdict poses no issue in terms of the PoI’s binary verdict requirement. I’m grateful to Ailbhe O’Loughlin for making me be clearer about this.
[3] Interestingly, this conception of the PoI as being grounded in preservation of the pre-trial status quo was itself set out in Federico Picinali, ‘The Presumption of Innocence: A Deflationary Account’ (2021) 84(4) MLR 708-739
Published 28.09.22
Response by Jackson Allen
DOWNLOAD PDF