A. Justificational Priority

The idea of justificational priority is concerned with the direction of justification. Some procedures are justified in virtue of their likelihood to bring about a just or correct outcome. Think, for example, of dividing a cake using the ‘you-cut-I-choose’ method: the one who cuts the cake is the last to get her share. This procedure is justified in virtue of its likelihood to lead to the just outcome—an equal division of the cake. However, with regard to other procedures, the direction of justification is reversed: the outcome is justified in virtue of it being a product of an independently justified procedure. A paradigmatic example is allocating an indivisible good between equally entitled claimants via a fair lottery, such as a coin toss. We refer to these two phenomena as ‘justificational priority’ of the outcome and or the procedure, respectively.^{Footnote 22}

Unlike the coin toss and cake division examples in which identifying the direction of justification is rather straightforward, the task is more complicated in the legal context, as often the justification criteria are a mix of substantive and procedural factors. However, there are cases in which the dominant direction can be identified. Consider, for instance, regulation that aims to gain the most valuable offer through public auction or competitive public bidding. Such procedural regulation is justified because it leads to a certain desirable outcome (e.g., maximizing public gain). By contrast, rules of estoppel, for example, reflect the opposite direction of justification. The concrete legal outcome is not justified independently of the procedure but rather in virtue of the fairness considerations underlying the procedural rule (of estoppel) that brings about that outcome.

Cases in which the procedure has justificational priority over all its consequences,^{Footnote 23} or the ultimate justification of the outcome rests upon its being the result of a certain procedure, can be referred to as cases of pure justificational priority of the procedure. In such cases, a deontological consideration must be involved, because consequentialist theories lack the resources to justify the procedure without referring to some consequences thereof.

Consider, for example, the right to make full answer and defence. On one account of this procedural right, the outcome has justificational priority over the procedure in the sense that this procedural right is justified in virtue of the instrumental contribution a hearing makes to the accuracy of the outcome of a trial. On another account, the procedure has justificational priority over the outcome because denying a defendant the opportunity to be heard is unjust, regardless of whether or not a hearing would assist in reaching the right verdict.^{Footnote 24} Yet, even according to this ‘procedural’ account, the right to be heard would not have pure justificational priority unless it reflects a deontological value. If it reflects a consequentialist consideration—albeit one intrinsic to the legal process itself as opposed to its outcome (e.g., if it is vindicated by the role that a hearing plays in creating a sense of self-respect)—the procedure does not have pure justificational priority.^{Footnote 25}

The notion of justificational priority can be easily conflated with another, somewhat close notion, which is concerned not with justification but with truth-making. A central question of philosophy of law and jurisprudence is whether the criterion for the validity of legal norms, or the truth maker of legal propositions (e.g., ‘the defendant is guilty as charged’), is procedural (e.g., ‘the defendant was convicted following due criminal process’) or substantive (e.g., ‘the defendant perpetrated the crime’).^{Footnote 26} Whenever the criterion for the truth of a legal proposition (or the validity of a legal norm) is procedural, we will refer to the procedure as having ‘metaphysical priority’ over the substance. This follows vice versa: whenever the criterion for the truth of a legal proposition (or the validity of a legal norm) is substantive, we will say that the substance has ‘metaphysical priority’ over the procedure.^{Footnote 27}

Procedures that have justificational priority may also have metaphysical priority. For example, a coin toss procedure used to determine which tennis player will serve first—Serena Williams or Maria Sharapova—has justificational priority over the outcome. It may also have metaphysical priority, since the criterion which determines the truth-value of the proposition that, say, Serena is the one who should serve first (i.e., the proposition’s truth maker) is procedural. The proposition is true if and only if Serena won the coin toss.

Generally speaking, the term ‘justificational’ may refer to different normative systems: epistemic, moral, legal, religious, and so forth. The distinction between justificational priority and metaphysical priority is clear when the direction of justification means the direction of epistemic justification. For example, perhaps what justifies our belief in Fermat’s Last Theorem is the fact that certain mathematicians have proved its correctness. However, this is not to say that the correctness of Fermat’s Theorem depends on the application of the proof procedure by the mathematicians.

Nevertheless, the distinction between justificational priority and metaphysical priority is less clear when it comes to moral or legal justification.^{Footnote 28} It is difficult to point to the exact difference between the proposition that the criterion for the moral justification of a certain action is procedural, and the proposition that the criterion for the truth or validity of the claim that the action is morally right is procedural. In these normative contexts, the distinction between justification and truth is often obscured.^{Footnote 29}

B. Normative Priority

Let us consider now a second priority relation between procedure and substance, which refers to the normative supremacy of procedural over substantive norms. We first present the general idea of a certain procedure having ‘normative priority’ over the substantive outcome. We then distinguish between two different scales and contexts in which the question of normative priority of procedural norms may arise.

A. Probative Evidence Seized in Violation of Rights

The classification of criminal evidentiary rules as a case of justificational priority of a certain outcome—such as the outcome of minimizing erroneous convictions while maintaining a reasonable frequency of erroneous acquittals—reflects a consequentialist approach, according to which the moral justification of these rules depends solely on the consequences of their application. This classification can be rejected on general moral grounds by denying the consequentialist account of the relevant evidentiary rule. The debate over the direction of justification—which is related, as mentioned, to the debate between consequentialist and deontological approaches to evidence rules—can be illustrated by the case of probative and reliable evidence that was seized in violation of rights, such as the right to privacy or autonomy.

Current trends in North American and English jurisprudence regarding the inadmissibility of unconstitutionally or illegally obtained evidence demonstrate the alleged departure from the lexical dichotomy approach. Albeit moving in opposite directions, different legal systems are driven towards a flexible, contextual ‘balancing approach.’ For example, in R v Grant the Supreme Court of Canada revised R v Stillman’s “all-but-automatic exclusionary rule for non-discoverable conscriptive” test, and adopted a more sensitive balancing test for exclusion which only takes this fact into account, among other factors, to assess the effect of admitting the evidence on society’s confidence in the justice system.^{Footnote 48} In the US, the exclusion of evidence seized in violation of constitutional rights was generally mandatory.^{Footnote 49} However, in the last decades, the scope of the Fourth Amendment exclusionary rule had been gradually narrowed down.^{Footnote 50} In contrast to the American constitutional tradition, the common law traditionally refused to consider the circumstances in which the evidence was obtained as a reason for inadmissibility. Yet, contemporary legislation and the gradual constitutionalization of English criminal procedure demonstrate a departure from the common law tradition. Most notably, section 78 of the Police and Criminal Evidence Act, 1984 provides courts with judicial discretion to exclude evidence if, given the circumstances under which it was seized, the admission of that evidence would have an adverse effect on the fairness of the proceedings. The concept of ‘fair trial’ as applied by the English courts is, however, highly flexible, open-ended, and context-sensitive.^{Footnote 51}

The justification for the exclusion of probative and reliable evidence seized in violation of rights cannot rely on the desirable outcome of sustaining a low rate of erroneous convictions. This is not to say, however, that such exclusionary rules cannot be justified in virtue of their leading to another outcome—for instance, guiding police behaviour in future cases and thereby minimizing violations of rights.^{Footnote 52} Notice that in this case, the exclusionary rule also does not display pure justificational priority, which requires that the justification of the rule be non-consequentialist. We turn now to discuss a more detailed illustration of this point.

Let us first stipulate that a certain exclusionary rule, which excludes probative and reliable evidence seized in violation of rights, does not have pure justificational priority—namely, it is justified in virtue of its instrumental role in bringing about desirable outcomes, such as minimizing future violations of rights.^{Footnote 53} Under this assumption, it seems that at least sometimes, such as in cases of very serious crimes, it is morally justified not to acquit a defendant who is guilty, even though his rights were violated.^{Footnote 54}

In this regard, there seems to be a gap between the ex-ante perspective—when the law aims to deter unlawful police conduct in order to minimize the violation of rights of the presumably innocent—and the ex-post perspective—when it is proven that the defendant has committed the crime. It might be argued that the fact that the defendant is guilty is relevant not only to the all-things-considered normative status of the legal outcome, but also to the normative status of the violation itself (the act which has violated the defendant’s rights). According to this suggestion, a state of affairs wherein an innocent person is subject to an unlawful search is morally worse than a state of affairs wherein a person who, for instance, was illegally carrying a gun is subject to the same unlawful search, even if there is no difference between the epistemic (mental) state of the relevant agent (i.e., the policeman) conducting the search in each of the two cases.

The existence of a gap between the ex-post and ex-ante perspectives is typical to substance-procedure dilemmas.^{Footnote 55} For instance, from an ex-ante perspective, the criminal procedure aims at providing incentives to submit a notice of appeal on time, but from an ex-post perspective, it might be wrong to reject a justified criminal appeal only because the appellant (or his lawyer) failed to submit it on time. The solution in this case might be to create significant incentives to submit an appeal on time without making the substantive outcome of the appeal strictly conditional on when the appeal was submitted. Back to the question of evidence seized in violation of rights, assuming that the only justification for an exclusionary rule is the consequentialist reason of guiding police behaviour in future cases, one could argue that a similar solution is called for: creating significant incentives to avoid the violation of rights without making the substantive outcome of trial strictly conditional on the legality or constitutionality of the (probative) evidence. Indeed, on the view that the rationale of the exclusionary rule is deterrence against rights’ violations, the justification of the acquittal outcome does not relate to the actual parties to the proceeding, and is in fact unjustified in the realm of the parties.

It seems, then, that if the only justification for excluding the evidence is deterring enforcement officials, there is room for allowing judicial discretion to deviate from the exclusionary rule in certain circumstances. In particular, if the end result of deterring unlawful police conduct can be achieved by means other than acquittal, it might be appropriate to deviate from the exclusionary rule.^{Footnote 56}

To summarize this point, assuming that the exclusionary rule does not have pure justificational priority, but rather is justified in virtue of its leading to a certain desirable outcome (such as minimizing future violations of rights), a deviation from that rule in certain circumstances might be appropriate.

A different conclusion may be reached if the exclusionary rule does have pure justificational priority over the outcome. An interesting question is (putting aside ‘external’ consequentialist reasons, such as deterring enforcement officials from violating rights or securing public confidence in the fairness of legal proceedings^{Footnote 57}), is it morally required to suppress illegally obtained (probative) evidence due to ‘internal’ reasons that touch on the realm of the actual parties to the criminal proceeding?

It could be argued that illegally obtained evidence should be excluded in order to vindicate the relevant rights, or that exclusion is the only appropriate remedy for such violation of rights—as opposed to the view that the defendant can be compensated for the violation of his rights in a different way than acquittal.^{Footnote 58} Intuitively, there seems to be something unfair about convicting a defendant on the basis of evidence that could not be obtained unless the police violated his rights, even if his guilt can be proven beyond a reasonable doubt. In such a case the defendant seems to have a kind of estoppel claim: the prosecution is ‘estopped’ from prosecuting on the basis of evidence that the state itself defines as unlawfully seized.^{Footnote 59} On this view, the exclusion of probative evidence seized in violation of rights is not justified by the consequences thereof—for example, in terms of minimizing future violations of rights or protecting public confidence in the legal procedure—but rather reflects a deontological constraint of fairness. This view, if sound, generates a counter-example to the proposition that evidence rules are always justified in virtue of their leading to desirable consequences, such as guaranteeing a low frequency of erroneous convictions or minimizing violations of rights.

It is doubtful, however, that such estoppel-like deontological consideration (even if valid) is absolute. A moderate deontological constraint—which can be defeated if the consequences of applying it are too harsh—seems more plausible than an absolute one. It could be argued, then, that exclusionary rules that exclude probative and reliable evidence seized in violation of rights (such as the right to privacy) are different than inadmissibility rules that exclude epistemically defective or misleading evidence.^{Footnote 60} The latter rules have pure justificational priority over the concrete outcome (which entails lexical normative priority), because the factual decision is epistemically justified solely in virtue of the epistemic status of the evidence. On the other hand, the former exclusionary rules that exclude probative and reliable evidence reflect only one consideration of justice that should be weighed against other considerations. That is to say, inadmissibility rules that exclude illegally obtained probative evidence do not display lexical normative priority over the outcome.^{Footnote 61}

In response, a proponent of the view that such exclusionary rules do have lexical normative priority could still insist that, even if the above analysis is correct with regard to the moral justification of the exclusion, it fails nonetheless with regard to the legal justification thereof. Thus, even if the convicting outcome is morally justified despite the illegality in obtaining the evidence, that outcome cannot be legally justified. This state of affairs could be compared to a case in which the judge herself saw the accused committing the crime. Even assuming that the judge is extremely reliable and thus epistemically justified in her (almost one hundred percent) certainty that the accused is guilty, a conviction on the basis of such personal knowledge of the judge is still obviously illegal, regardless of the existence of valid moral reasons against acquittal (e.g., protecting the public from a dangerous criminal).

Indeed, there might be cases in which an absolute deontological constraint can generate pure justificational priority of a procedural norm. The above example and the case of evidence obtained by torture might be such cases. However, a general absolute constraint against the admissibility of illegally obtained probative evidence seems morally and legally implausible. Even if there is a general fairness consideration against the admissibility of such evidence which is prima facie valid, there could be other normative considerations that would render the admissibility of illegally obtained evidence all-things-considered justified.

B. Statistical Evidence and Character Evidence

The hypothesis that criminal procedure in general, and criminal evidence rules in particular, cannot be explained solely by referring to the desirable consequences of their application—for example, in terms of accuracy in fact-finding, minimizing erroneous convictions, or minimizing violations of rights—receives support from the distinction between statistical and individual evidence, and the long-standing resistance of courts and legal scholars to use statistical evidence as a basis for criminal conviction.^{Footnote 62} If evidence norms were justified solely in virtue of their likelihood to bring about accurate outcomes, then even mere statistical evidence would presumably not be considered problematic. Indeed, in scientific disciplines such as medicine, reliance on statistical evidence is quite common, and such evidence is perceived as no less legitimate than individual evidence.

Consider, for example, the ‘gatecrasher case,’^{Footnote 63} where a stadium’s gate was crashed and many people entered the stadium without purchasing tickets. Even if it is known that, say, one thousand people attended the game and only ten of them purchased tickets, in the majority of legal systems it would not be possible to convict someone from the crowd based on the statistical evidence that 99 per cent of the audience had not purchased a ticket. In contrast, it is clearly permissible to convict someone from the crowd based on a 99 per cent reliable eyewitness.

The puzzle is why it is justified to distinguish between these two types of evidence, when the probabilities are exactly the same. If, from an epistemic point of view, there is no difference between statistical and individual evidence—or even if there is an epistemic difference between them (for instance, because statistical evidence cannot ground knowledge) but this difference is not legally relevant, since what is important in terms of accuracy in fact-finding are the probabilities (which are assumed to be exactly the same)^{Footnote 64}—then what can justify the different attitudes of the law towards these two types of evidence?

One recently suggested type of vindicating explanation is based on practical-instrumental considerations. It was argued that excluding statistical evidence may have better effects in terms of incentives. In a legal system in which statistical evidence is admissible, a person has less incentive to avoid criminal activity. This is because a court’s belief in his guilt, which is based on statistical evidence, is ‘insensitive’ to the truth:^{Footnote 65} had it not been the case that the defendant actually committed the offence, a court which bases its decision on statistical evidence would still find the defendant guilty. Returning to the gatecrasher case as an example, the potential offender’s decision of action would have almost no influence on the statistical evidence, and he knows that he will be convicted based on that evidence regardless of whether he purchases a ticket or not.^{Footnote 66} In contrast, a court’s belief based on eyewitness evidence is sensitive to the truth: had it not been the case that the defendant committed the offence (i.e., if he did not gatecrash) the court most probably would not have found him guilty on the basis of eyewitness testimony. Therefore, excluding statistical evidence and relying only on individual evidence creates more efficient incentives to avoid criminal activity.^{Footnote 67}

A similar incentive-based, instrumental explanation was suggested for the rule which defines character evidence as inadmissible for proof of conduct (as opposed to its admissibility in the sentencing stage). It was argued that since character evidence can be of probative value, the inadmissibility of such evidence cannot be explained if evidence law is perceived as exclusively aimed at finding the truth. However, a different kind of explanation, based on incentives, is available. The optimal incentive structure is the one that maximizes, from the agent’s perspective, the correlation between the actual decision to commit a crime and the likelihood of being convicted. Admissibility of character evidence reduces this correlation since the existence of that evidence does not depend on the decision of whether or not to commit the crime.^{Footnote 68}

The problem with such explanations is that even if they provide an adequate justification for the normative judgements against the admissibility of statistical and character evidence, they do not explain their phenomenology: the pre-theoretical intuitions against relying on such evidence. It seems that the intuition against relying on statistical and character evidence remains intact, even in a hypothetical world in which acoustic separation exists and people’s behaviour is not affected by evidentiary rules (which are selectively addressed only to the judicial and enforcement authorities). In such a hypothetical world of acoustic separation, evidentiary rules do not create any incentive, and therefore the incentive-based practical considerations for excluding statistical evidence and character evidence are no longer valid.^{Footnote 69}

If the epistemic and the practical-instrumental explanations of these evidentiary rules should be denied,^{Footnote 70} then what remains? Assuming that these rules are indeed justified,^{Footnote 71} we are left with a non-consequentialist explanation that is somehow related to a deontological requirement of ‘procedural fairness.’^{Footnote 72} If this is true, then perhaps we have here additional examples for pure justificational priority of procedural norms. As already discussed, pure justificational priority of a procedural norm implies lexical normative priority of that norm, which leaves no room for judicial discretion to deviate from it. The fact that in many legal systems the inadmissibility of character evidence at the trial stage has normative priority, and courts usually do not balance between the considerations in favour of their inadmissibility and other considerations,^{Footnote 73} such as accuracy in fact finding, is indeed compatible with the hypothesis that these inadmissibility rules reflect a deontological consideration of fairness rather than consequentialist, instrumental considerations.^{Footnote 74}

Similarly, the fact that in many legal systems defendants are rarely convicted on the basis of mere statistical evidence (of the type described in the gatecrasher example)^{Footnote 75} is more compatible with the view that this exclusionary rule reflects a deontological consideration rather than with the view that it reflects consequentialist (e.g., incentive-based) considerations. Indeed, if the distinction between statistical and individual evidence is vindicated by a practical-instrumental explanation, such as that relying on statistical evidence will render the behaviour incentives the law provides less efficient than they would otherwise be, then we should expect to encounter a somewhat ‘softer’ rule, that would balance between the consideration of creating efficient incentives and the consideration of accuracy in determining the facts (that supports the potential admissibility of any reliable evidence, statistical evidence included).

We thus contend that the fact that courts have traditionally refrained from weighing the considerations underlying the restrictions on the admissibility of statistical evidence and character evidence against other normative considerations (such as accuracy in fact-finding) lends some support to the hypothesis that these evidentiary rules are not vindicated by consequentialist (e.g., instrumental) explanations. The current trend of softening the traditional exclusionary rules against statistical and bad character evidence^{Footnote 76} can thus be considered as another example for the departure from the lexical dichotomy approach and for the modern trend to blur the distinction between substance and procedure.

C. The Reasonable Doubt Standard

The standard of proof beyond a reasonable doubt is a paradigmatic example of a procedural norm which is perceived to have global normative priority.^{Footnote 77} Notwithstanding this legal truism, developments in evidence law during the last century, associated with the tendency to relax the lexical dichotomy approach, might be considered as erosion of the global normative status of the reasonable doubt standard and evidentiary rules that are related to it, and accordingly as a derogation from the presumption of innocence.^{Footnote 78} In Canada, for instance, courts chipped away at common law exclusionary rules that used to be perceived as safeguards against conviction on the basis of problematic (unreliable, misleading, prejudicial, etc.) evidence. Examples are the elimination of corroboration requirements,^{Footnote 79} the relaxation of the hearsay rule,^{Footnote 80} and the gradual erosion of the constraint on the admission of character evidence.^{Footnote 81} Another example for the recent developments is the constantly increasing number of statutes which include reverse onus clauses that impose a burden on the defendant to disprove an element of the offence or to prove a statutory defence.^{Footnote 82}

The debate over the appropriate interpretation of the presumption of innocence demonstrates nicely the abovementioned conflict between the traditional lexical dichotomy approach and the modern trends which blur the distinction between substance and procedure. The dispute is between those who defend a restrictive procedural interpretation of the presumption of innocence—according to which it applies only to the elements of the offence^{Footnote 83}—and those who opt for a substantive interpretation—according to which courts should be allowed to review not only reverse burden presumptions and defences, but also the substantive content of criminal law such as offences (i.e., strict liability) which do not require proof of mens rea.^{Footnote 84} The former account considers the presumption of innocence to be purely procedural: it concerns merely the epistemic (uncertainty) threshold for criminal liability, which is independent of the content of the particular offence that needs to be proven. This account thus fits well with the view that ascribes lexical global normative priority to the reasonable doubt standard, and rejects the idea of a context-sensitive criminal standard of proof. In contrast, the substantive interpretation denies the sharp distinction between substance and procedure: it is not concerned (merely) with the epistemic threshold for conviction, but rather with the broader substantive issue of the moral wrongness of criminalizing persons who are not blameworthy. Accordingly, it does not find a real difference, in this respect, between convicting a person whose guilt was not proven beyond reasonable doubt and convicting a person in the absence of mens rea.^{Footnote 85}

If our concern is with the general issue of the moral wrongness involved in convicting and punishing persons who might be morally innocent, there seems to be no reason to refrain from considering any moral consideration which is relevant to the all-things-considered conclusion: the gravity of the crime, the severity of the potential punishment, the potential danger to the public safety, the psychological effect on the victim, and so on. In other words, the substantive account of the presumption of innocence fits well with a context-sensitive criminal standard of proof—namely, with the view that the reasonable doubt standard does not have lexical global normative priority over the substance. In this respect, this account has the potential to pull in opposite directions: on the one hand, broadening the protection against morally wrong conviction by allowing judicial scrutiny of the content of criminal offences, but on the other hand, providing courts with the discretion to consider any moral consideration which is relevant to the dilemma about the degree of certainty required for criminal conviction and balance between the conflicting considerations.^{Footnote 86}

It could be argued further that the derogation from the presumption of innocence is reflected not only in evidentiary rules that refer directly to burdens of proof, such as reverse onus clauses, but also in the constant softening of other traditional evidentiary rules. A possible argument for this claim is the following. An adequate meaning of the reasonable doubt standard of proof cannot refer merely to the fact-finder’s high subjective probability (or confidence) about the defendant’s guilt. This is because the firmness of a belief does not entail whether that belief is rational or well founded on the evidence. Therefore, a more objective sense of ‘beyond reasonable doubt,’ which is not in terms of the fact-finder’s mental state, is required.^{Footnote 87} Such interpretation of the reasonable doubt standard may require a rational high degree of belief in the defendant’s guilt. The traditional common law admissibility rules might be perceived as safeguards against conviction on the basis of problematic (unreliable, misleading, prejudicial, etc.) evidence, such as bad character evidence. The considerable relaxation of these admissibility constraints, the argument concludes, allows conviction on the basis of evidence that is not sufficient for granting a rational high confidence about the defendant’s guilt.^{Footnote 88}

This argument gets further support from empirical studies that show that there are substantial differences in the way different fact-finders perceive and apply the concept of ‘reasonable doubt.’^{Footnote 89} These findings suggest that the outcomes of criminal procedure which eliminates evidentiary constraints and guides fact-finders only by asserting that a defendant should not be convicted unless found guilty beyond a reasonable doubt heavily depend on who is the fact-finder, and therefore such criminal procedure is highly unreliable.^{Footnote 90}

The abandonment of the ‘one-size-fits-all’ approach in favour of a more flexible approach to evidence law also made room for a dependency on substance, as it allows courts to accommodate the particular content of evidentiary norms to the specific substantive legal context and the particular circumstances of the case. These trends allowed legislature and courts to weigh the considerations that lie behind the traditional evidence rules against other policy considerations that are related to the particular substantive matter—the severity of the crime, the unavailability of other evidence, the victim’s right to privacy, and so on.^{Footnote 91}

It is disputed whether this flexible approach in fact erodes the reasonable doubt standard and the ability of the accused to make full answer and defense^{Footnote 92} or if rather it is nothing more than being true to the rationales of evidence law (including the presumption of innocence) and sensitive to the circumstances of the particular case.^{Footnote 93} At a more fundamental, moral level, the question is whether the standard of proof beyond a reasonable doubt should have global priority. Should the same standard apply uniformly to all criminal offences, or should it depend on the type of the offence, and take into consideration the seriousness of the offence and the degree to which offences of its type are difficult to prove? This formulation of the question presupposes that the reasonable doubt standard is a context-independent epistemic standard. Put differently, the question is whether the legal concept of ‘reasonable doubt’ should have a context-dependent content.^{Footnote 94} Full discussion of this normative question is beyond the scope of this article. Nevertheless, let us make the following general remark.

From the perspective of a consequentialist moral theory, it might be justified to deviate from the standard of proof beyond a reasonable doubt (assuming now that it is context-free) in certain circumstances.^{Footnote 95} For example, in cases of serious-but-hard-to-prove crimes, the consequences of applying the reasonable doubt standard of proof (e.g., in terms of social utility) might be worse than the consequences of applying a lower standard.^{Footnote 96} This is so because a defendant who committed a serious crime might create a risk to the public if mistakenly acquitted that exceeds the risk that would be created by an acquittal of a defendant who committed a minor crime. In addition, the harder it is to prove an offence, the greater the probability of acquittals of blameworthy (and presumably dangerous) defendants.

This conclusion is mitigated, however, by the fact that a wrongful conviction for a serious offence is costlier (in terms of disutility to the convicted person) than a conviction for a less severe offence. In particular, if we take into account also the dependence of the punishment on the severity of the crime (assuming it is fixed),^{Footnote 97} there might be circumstances in which the severity of the crime (and, accordingly, the punishment) would be a reason for raising the standard of proof. Similarly, in case of minor crimes that carry mild punishments, the cost of mistaken conviction is low and this should count in favor of applying a lower standard of proof.^{Footnote 98}

It seems therefore that the global character of the reasonable doubt standard is more compatible with a deontological moral theory than with a simple consequentialist theory.^{Footnote 99} Recalling the connection between pure justificational priority of procedure and the existence of deontological consideration, this conclusion stands in line with the view that the procedure has justificational priority over the concrete outcome of a criminal trial, namely, a conviction of a specific defendant is justified in virtue of the fact that he was convicted under a fair and reliable criminal process.

In principle, a departure from the reasonable doubt standard can be justified also from the perspective of a moderate deontology, according to which deontological constraints can be overridden if the ‘good’ resulted (or the ‘bad’ avoided) by infringing it exceeds a certain threshold. In cases in which the consequences of applying the reasonable doubt standard are extremely bad, that threshold might be met.

One may ask whether it is more plausible to understand the reasonable doubt standard as reflecting an absolute deontological constraint or a moderate one. Seemingly, it is more plausible to interpret this standard of proof as reflecting a moderate deontological constraint, since an absolute constraint on convicting innocents would arguably necessitate the acquittal of each and every defendant. However, it can also be argued that the absolute constraint is not against convicting innocents, but about convicting a person (or, perhaps, an innocent person) whose guilt was not proved beyond a reasonable doubt. That is, every person (or perhaps every innocent person) has a right to be protected against conviction if his guilt was not proved beyond a reasonable doubt. If so, the standard of proof should not be affected by the seriousness of the crime.

Indeed, from a deontological point of view, there is a need to distinguish between the ‘frequency threshold’ and the ‘probability threshold’ for convicting innocent persons, meaning that there is a normative difference between a state of affairs in which for achieving a certain good outcome, ninety-nine guilty defendants and one defendant who is certainly innocent are convicted; and a different state of affairs, in which for achieving the same outcome, a hundred defendants whose probability of innocence is that 0.01 are convicted.^{Footnote 100} We thus have here an additional counter-example to the claim that the standard of proof beyond a reasonable doubt is normatively justified in virtue of the fact that it leads to an outcome of low frequency of wrongful convictions, for the relative frequency of wrongful convictions is identical (1 per cent) in both cases.^{Footnote 101}

The lack of equivalence between the frequency threshold, which is concerned merely with consequences, and the probability threshold means that one can coherently hold an absolute deontological position with regard to the reasonable doubt standard. However, it is still unclear what the rationale of such an absolute deontological constraint is. Arguably, the justification for occasionally convicting an innocent defendant (who was found guilty beyond a reasonable doubt) is related, one way or another, to the severe outcomes of demanding an absolute certainty for conviction. It seems, then, that the best account of the reasonable doubt standard is in terms of moderate deontology, and therefore it may be permissible, in principle, to deviate from this standard in cases in which the consequences of adhering to it are extremely harsh.