Abstract
This article tackles the question: can the Presumption of Innocence (PoI) be a presumption? Whereas many criminal law theorists rejection such a notion, I draw inspiration from argumentation theorists and philosophers—in particular, Petar Bodlović and Edna Ullmann-Margalit—and argue in favour of it; indeed, argumentation theory often holds the PoI out as a paradigmatic presumption. My argument proceeds in three sections. I first show that criminal law theorists writing on the PoI have understood presumptions as evidentiary devices in the form of a modus ponens. On that understanding, the PoI cannot be a presumption. Attention is then drawn to the field of argumentation theory, which teaches us that there are other types of presumptions that are non-evidentiary, not in the form of a modus ponens, require a tentative commitment to q, and require an agent to proceed (act) as if q; viz practical presumptions. The PoI can be understood as such. Finally, it is argued that the PoI, insofar as it requires a tentative commitment to q (here, ‘the defendant is innocent’), can be thought of as a propositional imagining of q (ie, an agent presuming innocence is to propositionally imagine the defendant's innocence).
Keywords
Introduction
Many legal writers argue that the presumption of innocence is not an authentic presumption (Gama, 2017: 559)
[T]he presumption of innocence … is considered a paradigmatic example of presumption among most argumentation scholars (Bodlović, 2017: 516)
Most criminal law theorists baulk at the idea that the Presumption of Innocence (PoI) is an evidentiary presumption. Some go further and accordingly claim it is not a presumption at all (Gama, 2017: 559). Those who hold this latter view I shall call Sceptics. 1 On such a view, the PoI is not a presumption in any sense of the defendant's innocence: ‘presumably, the defendant is innocent’ is absent from the subject's reasoning. My aim is to expose the false assumptions that underpin the Sceptical view, and to show that the PoI can be understood as a true presumption.
The first part outlines and responds to the false assumptions made by many Sceptics. The first of these equates a presumption of innocence with a belief in innocence (hereafter, the Belief Thesis). This equivalence gives rise to two seemingly fatal problems (addressed under ‘The Belief Thesis’): it is thought that a presumption—conceived as a belief—in a defendant's innocence would stymy the investigation and prosecution of criminal offences (Ulväng, 2013: 218); and it is thought that a presumption—conceived as a belief—is not voluntary and thus not a propositional attitude that jurors (or anybody else) can adopt on instruction (Picinali, 2021: 727; Roberts, 2020: 8916).
The second false assumption is exposed by argumentation theory. Many Sceptics have refused to see that the PoI is a presumption because they have latched onto a narrow, evidentiary understanding of ‘presumption’ as a factual inference in the form, ‘if p then pres q’ (similar to the evidentiary presumption that a person is presumed deceased if she has been missing for over seven years) (for example, Weigend, 2013: 193–194). Argumentation theory shows us that presumptions are a heterogenous device (Bodlović, 2019, 2020; Gama, 2017; Godden and Walton, 2007; Lewiński 2017), and that the PoI can be seen as a non-evidentiary type of presumption (but a presumption nonetheless)—what Nicholas Rescher (2006: 29) and Petar Bodlović (2021) term a ‘practical presumption’. 2 Benefitting from argumentation theory's more nuanced understanding of presumptions, I demonstrate that the PoI can be a (type of) presumption—specifically, a practical presumption with a non-evidentiary goal: of avoiding the greater harm of wrongfully convicting the innocent as opposed to falsely acquitting the guilty (Bodlović, 2019: 582, 2020; Ullmann-Margalit, 1983: 155). If I am right, we can ourselves be sceptical about the Sceptics’ reasons as to why the PoI cannot be a true presumption.
Having clarified the presumptive status of the PoI, some more positive claims follow (under ‘Presumption as a propositional attitude’). If not a belief, what propositional attitude is involved in this practical presumption underlying the PoI? First, I draw inspiration from Federico Picinali (2021), who suggests that presuming innocence might just be to propositionally imagine innocence. 3 I then consider an alternative characterisation of the PoI as a supposition of the defendant's innocence (legal presumptions are sometimes described in terms of suppositions). I conclude by siding with propositional imaginings on account of their motivational power.
My overall account of the PoI, insofar as it can be a presumption, 4 is that it is a practical presumption which requires a subject to propositionally imagine the defendant to be innocent and proceed as such.
False assumptions
The Belief Thesis
If conceived of as a presumption, the PoI might seem paradoxical: how can the police or prosecution presume the innocence of the defendant but also suspect her to be ‘someone in relation to whom a reasonable presumption of guilt exists’ (De Jong and Van Lent, 2016: 34; Mickonytė, 2019: 16)? If the police investigate and the prosecution prosecute, they do so (in part) 5 because the evidential weight justifies a reasonable presumption of guilt in the defendant (ie, there is a realistic prospect of conviction) (Packer, 1964: 11–12). As Aiste Mickonytė (2019: 16) points out, that reasonable presumption of guilt tends to be well-founded: ‘the experience of liberal democratic countries shows that the large majority of those formally accused of a crime are found to be actually guilty’. 6 Given this reasonable presumption of guilt is integral to a functioning system of criminal justice, and given there is (usually) a good empirical basis for this presumption, can presuming innocence and reasonably presuming guilt sit happily side-by-side? The Sceptic's worry is that they cannot, and if public authorities such as the police or prosecutors literally presume the innocence of suspects they will be ‘unable to do their jobs’ (Lippke, 2016: 24–25). This leads some scholars to contend that the PoI is not a presumption. Magnus Ulväng (2013: 217–218), for example, writes that when public authorities act pursuant to the PoI, they are ‘not necessarily presuming innocence, but rather acknowledging the possibility that [they] might be wrong’, for to presume innocence would be ludicrous—‘[o]therwise the police or a prosecutor would be committing an offence merely by initiating legal proceedings.’
Even if a reasonable suspicion in a person's guilt were consistent with a presumption of her innocence, an instruction to presume—in the sense of believing in—the defendant's innocence is not possible to comply with. 7 Picinali (2021: 20), for example, refutes the notion that the PoI can be a presumption of (a belief in) the defendant's innocence because ‘a requirement that someone believe something is fundamentally defective, since beliefs are not the sort of attitude that can be adopted at will’ (Picinali, 2021: 20). The impossibility of willing a belief into existence is, as Paul Roberts (2020: 8916) wonderfully remarks, like ‘Mogadon or mental kryptonite’ to this conception of the PoI.
Both objections to the Belief Thesis are sound. It is generally accepted that inconsistent beliefs cannot be held simultaneously (Foley, 1986, 1979), and that beliefs cannot be voluntarily adopted (Picinali, 2021: 727; Roberts, 2020: 8916). But the breadth of such objections should not be overstated; they only count against the Belief Thesis. That the PoI is a presumption remains a plausible view, so long as the ‘presumption of’ innocence is not understood to involve a ‘belief in’ innocence. Indeed, as will be later argued, presumptions that-q are just tentative commitments to q, which are to be distinguished from beliefs that-q (ie, doxastic commitments to q) (but see Wreen, 2003: 374). 8 Once we disavow the Belief Thesis and consider how the PoI could involve a propositional attitude other than belief, we can begin to salvage the claim that the PoI is a presumption. That task will be undertaken below.
Presumptions as homogenous, legal devices
The second false assumption made by Sceptics is more general. Whereas the Belief Thesis dooms only a specific way of presuming-q (a belief that-q), the second claim by Sceptics doubts whether presuming-q is any part of what the PoI requires (irrespective of that entailing a belief or some other propositional attitude). This objection holds that the PoI is not a presumption that-q because a presumption is restrictively understood as an evidentiary device in the form of a defeasible modus ponens: ‘if p then pres q’. This restrictive understanding of presumptions explains the Sceptic's view: the PoI is not a ‘true’ presumption because it does not conform to this conception of a presumption.
But the neighbouring field of argumentation theory teaches us there is no single, unitary account of presumptions. Rather, there are non-evidentiary and non-modus ponens types of presumptions. Perhaps the PoI can be one of these types of presumption.
Criminal law understandings of presumptions as defeasible modus ponens
Presumptions are by no means a settled concept in legal writings (Ashford and Risinger, 1969: 165; Gama, 2017; Lewiński, 2017: 596; Stuckenberg, 2014: 305; Walton, 2014: 85). Notwithstanding their differences, most legal theories of presumptions agree that a ‘minimal account [of] its basic structure’ (Walton, 2014: 85) is in the form of a defeasible modus ponens: ‘Most are agreed that a presumption is a legal mechanism which, unless sufficient evidence is introduced to render the presumption inoperative, deems one fact to be true when the truth of another fact has been established’ (Ashford and Risinger, 1969: 165). In other words, the bare skeleton of a legal presumption follows the form: If p (the fact proved), then q (the fact presumed).
9
It is because the fact-finder must know whether q in order to commence 10 or continue 11 a deliberative task, and because q is manifestly obvious but evidentially uncertain, that legal presumptions have purchase. By presuming q they can allow the fact-finder to ‘avoid wasting time and energy on claims that nobody would dispute’ (Walton, 2008: 227) (thereby grounding an initial premise from which deliberation may commence) or ‘bridge epistemic gaps’ (Roberts, 2020: 8921) (thereby allowing deliberation to continue where it has become stuck). Paradigmatic examples of such legal presumptions are the presumption in English law that a defendant knew or believed property to be stolen if it is proved that the property had recently been stolen, 12 or that a person is deceased if she has been missing for over seven years. 13 Presumptions of these sorts provide an ‘inference ticket’ (Wreen, 2003: 375) from p to q because ‘[o]ur general knowledge taken from the world of reasoning, science, and experience enable us to make a logical connection between two facts’ (Kitai, 2002: 264).
Against this legal model of presumption, two things about the PoI are worth mentioning. First, a modus ponens seems ill-fitting for the PoI—if innocence is the fact presumed (q), what is the fact proved (p) from which q derives? If charged with a criminal offence, then presumed innocent? 14
This is unconvincing. Being charged with a criminal offence can hardly be thought of as a ‘fact proven’ befitting of p.
15
Secondly, common experience and reasoning does not tell us that a person charged with a criminal offence is likely to be innocent (Weigend, 2013: 193–194). It tells us the opposite: ‘in reality, most individuals reported to the police for having committed a criminal offence are in fact guilty, not innocent’ (Mickonytė, 2019: 16; Weigend, 2014: 287). This second point is worth emphasising again, as it is the main argumentative leg on which Sceptics stand when they say the PoI is not a presumption. For example, Thomas Weigend (2013: 194) writes: The presumption of innocence … is not based on a certain set of facts, and it is not derived from general experience—certainly the fact that someone has been accused of a crime does not typically indicate that he has not committed that crime … [T]he presumption of innocence clearly differs from other (true) presumptions in that it has no basis in fact or human experience.
That Sceptics such as Weigend rely on this reason illustrates how they understand the nature of the ‘inference ticket’ in a modus ponens—that it is probability or ‘most plausibility’ (Rescher, 2003: 81–100) 16 connecting p to q. 17
The Sceptic's understanding of presumption as probabilistic and evidentiary
Emerging now is the following picture of ‘true’ presumptions, as used by Sceptics writing on the PoI:
Legal presumptions take the form of a defeasible modus ponens (if p, then pres q); and The inference ticket that permits the legal presumption that-q, from the proven fact, p, is probability or ‘most plausibility’: given p, our common sense and experience dictate that it is probable or most plausible that q, and so in the absence of proof that q, the law presumes q. The PoI is not a ‘true’ presumption because when someone is charged with a criminal offence, common sense and experience tell us that it is probable or most plausible that they are guilty (ie, improbable that they are innocent). Given A and B, we can see why Sceptics claim that: Legal presumptions are just ‘evidentiary’ presumptions: because common sense and experience tell us it is probable or most plausible that q (given p), the law takes p to be evidence of q, pro tem, in the absence of proof that q (or q).
Although Sceptics do not endorse B in terms as explicit as they do A, B can be negatively inferred from their arguments to the effect that:
Pursuant to this picture of ‘true’ legal presumptions, the PoI is not a presumption because it satisfies neither A, B nor D. And because it is not a ‘true’ presumption, the PoI does not require its subjects to presume q—in the sense of holding a ‘representational state with the content’ q.
18
By engaging with argumentation theory in the next section, I will demonstrate that presumptions need not be limited by A, B or D. Hence, the Sceptic's claim that the PoI is not a presumption is unwarranted.
Borrowing from argumentation theory
In reply to the Sceptical view, Antony Duff (2013: 269–270) accepts that a PoI does not satisfy A, B or D. Nevertheless, he says:
19
The PoI is not that kind of presumption: it specifies not a conclusion that the court may reach given certain evidence, but the position from which it must start. It is nonetheless worth calling it a presumption: for it specifies a legally relevant proposition, that the defendant is innocent of the offence charged, that the court must take for granted, and that proposition is usually not one that we would be empirically justified in taking for granted.
Duff is correct that the PoI can still be called a presumption. We have a defeasible proposition—that the defendant is innocent—that a public authority must adopt on a pro tem basis, in the absence of proof. Prima facie, the makings of a defeasible presumption seem present.
Enter argumentation theory
The study of presumptions is not only the bailiwick of evidence law. Presumptions are, as James Bradley Thayer (1898: 315) observed, just as important to the field of argumentation and legal reasoning: Presumptions, assumptions, taking for granted, are simply so many names for an act or process which aids and shortens inquiry and argument. These terms relate to the whole field of argument, whenever and by whomsoever conducted; and also to the whole field of law, in so far as it has been shaped or is being shaped by processes of reasoning.
With these words, Thayer whispered the last enchantments of the Victorian Age—he was, as Raymundo Gama (2017: 562) writes, ‘probably the last jurist who clearly saw that presumptions belong to argumentation … [and since then] Thayer's view on the place of presumptions has not been followed by contemporary legal writers’. Nowadays, legal scholars writing on presumptions tend to place presumptions within the exclusive domain of evidence law, and as such they have been largely unaware of the connection between presumptions in legal and non-legal argumentation … In contrast, scholars from other disciplines are fully aware that this subject [presumptions] goes beyond the legal domain, encompassing different disciplines.
Gama's charge of parochialism (2017: 562) against contemporary legal writings on presumptions is warranted. It helps explain why legal scholars so confidently say the PoI is not a presumption, whereas argumentation scholars say otherwise.
In argumentation theory, presumptions escape concrete definition (Gama, 2017: 564; Godden and Walton, 2007: 313–314, 337). One reason is that ‘studies of presumption in argumentation theory involve different theoretical goals, sources of inspiration, and, accordingly, different methodologies’ (Bodlović, 2019: 582). But we need not explore all the definitions of presumption and argue for a ‘correct’ one, nor do we need to identify the ‘hard core’ of a presumption in argumentation theory. It is enough for our purposes to recognise that presumptions are a ‘heterogenous’ (Godden and Walton, 2007: 333) and ‘pluralistic’ (Gama, 2017) device in argumentation—there are different types of presumptions, with different formal structures (Gama, 2017: 568; Stuckenberg, 2014: 135), and with different aims. Rescher (2006: 27) suggests that presumptions can be broadly split into two types: Presumptions obtain principally with two ends in view. On the one hand there are the purely cognitive presumptions made for the sake of answering our questions and filling gaps in our information (as, for example, presumptions regarding the reliability of sources). On the other hand there are practical presumptions made for the sake of guiding our decisions regarding actions … (Rescher, 2006: 27)
Later, Bodlović (2020: 257) accepted and expanded on this distinction: [W]e might distinguish cognitive (epistemic) and practical presumptions since they operate in different dialogical contexts (epistemic inquiry vs practical deliberation), perform distinct normative functions (promoting epistemic vs non-epistemic goals) and, thereby, have qualitatively different foundations.
In the next few subsections, I want to seize on this ‘cognitive-practical’ dichotomy. First, I contend that Sceptics—when claiming that the PoI is not a presumption—have solely understood presumptions as cognitive presumptions, without recognising the possibility of their being practical presumptions (see part iv). And second, retreating from this narrow, legalistic understanding of presumptions, and embracing a broader category of presumptions—like practical presumptions—can help show why the PoI is a type of presumption (just not one that Sceptics—looking only within the domain of evidence law—have recognised, at least explicitly) (see parts v–vii).
The Sceptic's model of presumption as a cognitive presumption
Recall the basic thrust of Part I(B)(ii): Sceptics view ‘true’ presumptions as an evidentiary device insofar as it tells us that the presumed fact, q, is probable or most plausible, given the fact proved, p. Put differently, the existence of p is taken to be evidence of q, pro tem. The PoI, understood thus, is not a ‘true’ presumption—the fact of being charged is not evidence of the defendant's innocence because, quite oppositely, it is probable or most plausible that they are guilty. This legal model of ‘true’ presumptions, I demonstrate in this subsection, is identical to a specific type of presumption in argumentation theory, viz a cognitive presumption.
In so limiting their conception of ‘true’ presumptions, Sceptics see only part of the picture. By this, I mean when Sceptics conceive ‘true’ presumptions as cognitive presumptions, they underappreciate other species of presumptions recognised in argumentation theory as having an equal (or at least competing) claim to being ‘true’ presumptions. I have in mind, specifically, practical presumptions. Before arguing that the PoI can be understood as a ‘true’ presumption qua practical presumption, I will first make good the claim that the Sceptic's model of presumptions and cognitive presumptions are one-and-the-same, or at least very similar, insofar as they are both about acquiring truths on the basis of evidence.
Cognitive presumptions are just those epistemic presumptions that are part of our cognitive toolkit in figuring out the way things are and acquiring true, justified beliefs about them (Bodlović, 2020: 266, 2021: 289–290). As Bodlović (2020: 258) argues, cognitive presumptions are ‘typically used in the context of [epistemic] inquiry’, in the promotion of epistemic goals (ie, finding out truth) (Rescher, 2006: 71). As such, in cognitive presumptions the thing presumed, q, is what Rescher (2006: 71) defines as the ‘most plausible candidate for truth’ that tentatively bridges a gap in epistemic inquiry until proven otherwise. Q being the ‘most plausible candidate for truth’, in turn, allows an agent to form a presumption that-q precisely because it is the ‘most plausible [truth] of rival alternatives’ (Rescher, 2006: 39). Two examples suffice to demonstrate cognitive presumptions in action. First is the presumption that the Earth is (roughly) round. Where the question is ‘what shape is the Earth?’ (an epistemic inquiry), science points to an overwhelming body of evidence showing Earth to be round. Equipped with that evidence, rational agents can then make the cognitive presumption that ‘the Earth is round’ because it is the ‘most plausible’ candidate for truth.
Second, consider a different epistemic inquiry where the question is ‘Where is Andy's cat?’ When Andy (who has lost his pet cat) looks outside his window and sees a cat in the tree that looks identical to his own missing cat, he forms a presumptive view that this cat is his: ‘although ‘Andy's cat is in the tree’ is uncertain from a sceptical viewpoint, Andy acts on it by relying on his senses and by using it as a tentative premise in reasoning’ (Bodlović, 2021: 289). In each example, particular attention should be given to how q, being the ‘most plausible truth candidate’, is supported by evidence: the mainstream scientific opinion is evidence of ‘the Earth is round’, whereas Andy's visual perception and memory are evidence of ‘Andy's cat is in the tree’(Bodlović, 2021: 301; Rescher, 2003: 81–100). See now that cognitive presumptions that-q are evidentiary in the same way that Sceptics hold ‘true’ presumptions to be: 20 in both cases, ‘a standard epistemic source … such as sense-perception, memory, testimony, expert-testimony, or common knowledge’ (Bodlović, 2021: 301) provides the evidentiary grounds for q as the ‘most plausible truth candidate’ (thereby enabling a cognitive presumption that-q).
With a basic picture of cognitive presumptions now sketched out, it is no wonder why Sceptics appear to deny the PoI of its presumptive force. ‘True’ presumptions, on the Sceptical view, must be cognitive—they should serve an epistemic goal in that the thing presumed, q, is the most plausible (Rescher, 2003: 81–100) or probable (Ullmann-Margalit, 1983: 157) ‘truth candidate’ about how things are (Rescher, 2003: 81–100) (in the absence of proof that q or q). As innocence is neither the most plausible nor probable ‘truth candidate’, 21 the Sceptic can confidently assert that the PoI is not a cognitive presumption; in fact, we presume innocence not in search of the truth, but to figure out what we should do given the unknown truth, and we presume innocence not on some evidential basis or epistemic source, but on the moral basis of preferring false acquittals over wrongful convictions. To the extent that Sceptics claim that the PoI is not a cognitive presumption, they are mostly right. 22 To the extent that Sceptics claim that the PoI is not a ‘true’ presumption, however, they are wrong.
The PoI as a practical presumption
The other half of the picture that Sceptics have neglected in their description of ‘true’ presumptions is found in practical presumptions. My claim here is that in argumentation theory, practical presumptions—as a species of presumption quite apart from cognitive presumptions—also have a claim to being a presumption, and therefore the Sceptic's understanding of ‘true’ presumptions is incomplete insofar as they are equated only with cognitive presumptions.
A brief explainer of practical presumptions is first necessary, whereafter I will show that the PoI squares with this type of presumption. The benefit of this approach is then exhorted.
Practical presumptions are goal-oriented; they are part of our practical toolkit in making decisions about what to do (ie, practical deliberation), rather than figuring out the way things are (ie, epistemic inquiry) (Bodlović, 2021: 289; Ullmann-Margalit, 1983: 143–144). A fortiori, practical presumptions do not lead to true, justified beliefs about the way things are, nor is the fact presumed ‘evidentiated’ from an epistemic source (Bodlović, 2021: 290): as Edna Ullmann-Margalit (1983: 296) wrote in her influential treatment of presumptions ‘[i]n a practical case, ‘presumably’ is, by definition, non-epistemic modality’. Rather than presuming to acquire truth, the thing presumed in a practical case is based on the goal of avoiding the greatest harm (Bodlović, 2021). Caught between two options—erroneously proceeding on q and erroneously proceeding on q—a practical presumption sides in favour of the safer, less costly option rather than the option that would lead to a true, justified belief that q or q. Notice that practical presumptions, in their natural habitat of practical deliberation (ie, deliberating over what we should do given q is unknown), is concerned primarily with actions rather than attitudes, and this accounts for why descriptions of practical presumptions all use the language of ‘proceeding as if’ (in the next subsection, I will defend the view that practical presumptions are not devoid in their concern with attitudes).
In view of practical presumption's non-epistemic and goal-oriented nature, Ullmann-Margalit writes that practical presumptions are less concerned with ‘chance of error’ (an epistemic consideration) as they are with the ‘acceptability of error’ (a practical consideration). Thus, in the case of a practical presumption, whether it is presumed q or q turns less on the probability or plausibility that q or q (a choice reflecting ‘chance of error’), but on which choice would produce a more acceptable type of error if erroneously proceeded upon: There is no question of avoiding errors; at best, there is the question of reducing their number. But a different sort of question is whether one type of error is to be preferred, on grounds of moral values or social goals, over the other(s). Evaluative considerations may exist which justify a systematic and generic bias in favour of erroneously proceeding on Q rather than erroneously proceeding on not-Q, given that P and given lack of sufficient reasons in the circumstances to believe either q or not-q to be the case. (Ullmann-Margalit, 1983: 159)
Though clearly not a cognitive presumption, the PoI now increasingly looks like a practical presumption. Delivering on this latter claim restores the presumptive status of the PoI. Like practical presumptions, the presumption in favour of innocence is not premised on the probability of innocence, nor is the presumption of innocence a tentative claim to truth or a ‘truth candidate’. Rather, the PoI, as with a practical presumption, is grounded in an important moral aim of avoiding the greater harm where q is uncertain and there is pressure to decide whether q: that it is worse to mistakenly punish the factually innocent than mistakenly absolve the factually guilty, and we presume in favour of the error that is less costly (in other words, it aims to avoid wrongful convictions), which are more morally repugnant than false acquittals.
Notice the explanatory contribution that argumentation theory offers here. Specifically, it can provide the structural framework for types of presumptions, and clarify the different function or goals of each type of presumption. To recap, cognitive presumptions are (1) tied to our cognitive faculties, notably beliefs; (2) about acquiring truth; and (3) based predominantly on evidence that shows q to be the ‘most plausible truth candidate’. Practical presumptions, on the other hand, are (4) non-epistemic; (5) about what to do; and (6) based on the goal of avoiding ‘greater harm in circumstances of uncertainty and pressure’ (Bodlović, 2021: 289–290). Whether any legal presumption truly is a presumption is, in my view, better answered by appealing to those frameworks developed by argumentation theorists rather than by appeals to the law of evidence.
The PoI fits into the framework of a practical presumption; all we need do is plug the relevant qualities of the PoI into that framework to test its presumptive status. Practical presumptions are non-epistemic, in the same way that the PoI lays no claim to truth or knowledge—requirement 4 is satisfied; practical presumptions are about what to do, in the same way that the PoI is about what we should do given a defendant has been charged but her guilt has not yet been determined—requirement 5 is satisfied; and practical presumptions are about avoiding greater harm, in the same way that the PoI is about avoiding wrongful convictions—requirement 6 is satisfied. The upshot of all this is that the PoI is a type of presumption.
Practical presumptions do include an attitude (a tentative commitment)
A great deal of emphasis has been placed on the argument that practical presumptions are about, inter alia, what to do (in other words, practical presumptions are about ‘proceeding as if-q’, ie, our actions). A legitimate objection could then be levelled against an account of the PoI that requires a ‘tentative commitment’—some type of non-doxastic, cognitive undertaking to q—beyond our actions. ‘As a practical presumption’, the objection would go, ‘the PoI governs only outward behaviour, and not our attitudes’. Bodlović (2021: 295) expresses (and ultimately rejects) this concern most clearly when he notes that if practical presumptions are really just about actions we should take to avoid the greater harm, there is no real need (in fact it might be dangerous) to take up any tentative commitment to q: ‘If the expected utility of [q] is greater than the utility of [q], then we can understand [the agent's] decision without referring to the concept of presumption. Introducing presumptions into practical reasoning, then, might be an unnecessary complication that leads to undesirable consequences: for instance, [the agent] becomes (needlessly) forced to include an epistemically unwarranted proposition, namely the presumption [q], into her commitment set.’
In the context of the PoI, the objection would proceed as follows: we can understand an agent's decision to proceed as if the defendant were innocent purely by reference to the aim of avoiding wrongful convictions, without any need for the agent to tentatively commit to the proposition ‘the defendant is innocent’ (which is epistemically unwarranted) to explain her actions.
However, such an objection can be countered. Although true that ‘the defendant is innocent’ is epistemically unwarranted, we are dealing here with a practical presumption, and thus ‘presumably, the defendant is innocent’ makes no epistemic claim. Presuming q in a practical, completely non-epistemic manner therefore does not lead to any ‘unnecessary complications’ (because q makes no claim to being epistemically warranted in the first place) nor ‘undesirable consequences’ (because any consequences in practical presumptions are the result of our actions, and not the result of our attitudes). The objection misses its mark and would only strike truly if the PoI were cognitive (a conception that I have refuted).
The objection also finds limited support in the literature. In contrast, there is considerable—though not unanimous—support for the view that practical presumptions do contain an attitude (a tentative commitment) to the presumed fact, q, on top of the ‘action policy’ of proceeding as if-q, so long as that attitude is not doxastic or epistemic (Ullmann-Margalit, 1983: 148–149). 23 Bodlović (2021: 296), for example, sees the continued relevance of an attitude (‘presumably, q’) in practical presumptions, not least because ‘presumably, q’ is ‘explanator[ily] relevant’ to why an agent proceeds as if-q. In other words, practical presumptions that-q involve both a rule of tentative cognition (‘presumably, q’) which explains the rule of action (to proceed as if-q)—all for the aim of avoiding greater harm.
Recasting the PoI as a practical presumption
Some basic conclusions now follow:
Sceptics take ‘true’ presumptions to be those presumptions satisfying A, B and D (see ‘ The Sceptic’s understanding of presumption as probabilistic and evidentiary’). Legal scholars’ understandings of presumptions have largely been insulated from interdisciplinary understandings of presumptions (such as those found in argumentation theory). Argumentation theory shows us that presumptions have a heterogenous nature—they do not always conform to the conditional structure of a modus ponens (thus A); they are not always probabilistic (thus B); and they are not purely epistemically driven (thus D). Argumentation theory shows us that presumptions can broadly be divided into cognitive presumptions and practical presumptions. Both are types of presumptions where an agent presumes q (ie, holds q in her mind and takes an attitude toward q). Sceptics tend to regard presumptions in the cognitive sense, without recognising that the PoI can be regarded as a presumption in the practical sense.
These conclusions cohere well with the literature on presumption in argumentation theory. They explain why Sceptics so often dismiss the PoI as a presumption, and why argumentation scholars treat the PoI as paradigmatic of presumptions. We can now say: the PoI is not really a cognitive presumption, but it can be understood as a practical presumption that requires both a tentative commitment that-q (‘presumably, q’) and acting as if-q.
Presumption as a propositional attitude
As I have just argued, far from being a ‘misnomer’ (Walton, 2014: 56), the PoI—as a practical presumption—does require its subject to presume innocence; that is, her reasoning must involve the tentative attitude ‘presumably, the defendant is innocent’. But what kind of attitude is involved in the modal operator, ‘presumably’?
Some attempts have been made between argumentation scholars and the few criminal law theorists who do conceive the PoI as a presumption. In argumentation theory, to presume q is commonly glossed over as a ‘tentative commitment’ to q (Bodlović, 2021: 291) or a ‘taking for granted’ that q (Rescher, 2006: 6; Ullmann-Margalit, 1983: 143; Wreen, 2003: 372). In Richard Lippke's understanding of the PoI, ‘presumably’ means to take up a ‘rebuttable perspective’ (2016: 100), ‘viewpoint’ (2016: 101), or ‘counter-intuitive stance’ (2016: 97) in relation to q. I find such periphrases to be unproductive—what is a ‘taking for granted’, ‘tentative commitment’, or a ‘counter-intuitive stance’, for example? And by framing the PoI as such, do we encounter the same difficulties raised by the objection to the Belief Thesis?
A useful place to start looking for answers is in the field of cognitive psychology. From this perspective, we can consider some potential cognitive processes that are constitutive of presuming, narrowing our focus to those cognitive processes that avoid the twin problems presented by the Belief Thesis; this means our focus is on those attitudes that meet, at least, two basic conditions:
And
Propositional imaginings
An answer might be found in our capacity to imagine q: essentially, this suggests that a presumption of the defendant's innocence is an imagination of the defendant's innocence.
To be clear, imagination is not a monolith. Amy Kind, for example, doubts we can speak of ‘the phenomenon of imagining’ (2013: 141) and points to a ‘philosophical literature … rife with distinctions drawn between different kinds of imagination’ (2013: 143). 24 In that light, when I suggest a presumption that q might just be an imagination that q, I am referring specifically to propositional imaginings. Importantly, to propositionally imagine q is just to take an attitude—imagining that—in relation to the propositional content q, 25 in a similar way that a belief or desire that q is to take an attitude of belief or desire toward q. As such, propositional imaginings are attitudinal and not imagistic or sensory (Kind, 2016: 5), and so do not involve the formation of mental imagery that other, non-serious types of imagination might. 26
At first blush, propositionally imagining innocence seems to satisfy the Belief Consistency and Voluntariness conditions: we commonly propositionally imagine things we believe not to be true (indeed, this is sometimes the very reason why we imagine), and we can generally initiate, redirect, and cease our propositional imaginations on command (Picinali, 2021: 728). That these qualities of propositional imaginings are widely accepted by philosophers make propositional imaginings an attractive answer to the question ‘what is it to presume innocence’?
But the attractiveness of characterising the presumption of innocence as a propositional imagining turns on a specific account of propositional imaginings: a representational account. To address this point, it is convenient here to take stock and remind ourselves that the PoI is a representational concept and therefore propositional imaginings must also be representational if they are to characterise the PoI.
A representational account of the PoI
So far, the argument has been that the PoI involves an actual presumption of the defendant's innocence, in the sense of presuming (taking an attitude toward) the defendant's innocence (the propositional content, q). Another way of expressing this is that the PoI involves ‘representationalism’, whereby an attitude is taken vis-à-vis a representation, q: ‘presuming’ being the attitude and ‘the defendant is innocent’ being the representation, q.
If, as some have argued, 27 propositionally imagining q does not involve holding the representation, q, in one's mind and taking an attitude toward it, it would seem to fail as an explication of any presumption that-q. It is not my intention here to settle the question of whether propositional imaginings ultimately do require one to hold the representational content q in one's mind; rather, it is only to point out that propositional imaginings—as a way of understanding presumptions—only have purchase if we latch onto a representational account of it. Because of this more modest goal, I will briefly sketch out a widely accepted account of propositional imaginings which fits the representational structure of the PoI (noting that non-representational accounts of propositional imaginings would be inapt to describe the PoI). 28
A representational account of propositional imaginings: Nichols and Stich's Possible Worlds Box
One of the most influential accounts of propositional imaginings is developed by Shaun Nichols and Stephen Stich (2000). At the crux of their theory is that in addition to beliefs and desires, which are foundational propositional attitudes that human beings take toward any given propositional content, q, 29 there is a third type of propositional attitude—imagination. Their account is often described as ‘boxological’ (Kind, 2016: 3), whereby each propositional attitude can be visualised in the cognitive architecture as boxes where particular representations or propositions can be stored—when we believe q, we store q in our Belief Box, or if we desire q, we store q in our Desire Box (Nichols and Stich, 2000: 121).
For our purposes, Nichols and Stich (2000) propose that when we propositionally imagine q, that representation token is stored in a separate box, the Possible Worlds Box (PWB). 30 When a proposition is placed in the PWB, I shall refer to this as an act of Imagining.
In this PWB are stored representation tokens that neither ‘represent the world as it is or as we’d like it to be, but … represent what the world would be like given some set of assumptions that we may neither believe to be true nor want to be true’ (Ichino, 2019: 1518; Nichols and Stich, 122; but see Sinhababu, 2013). On this account of propositional imaginings, we do hold the representational content q in our mind (in the PWB, specifically) and take an attitude toward it (Imagining). And because we hold the content q in our mind, this account of propositional imaginings conforms to the representational structure of presuming q, which also requires the holding of q (in this case, ‘the defendant is innocent’) in one's mind and taking a presumptive attitude toward it (in this case, an Imagining).
Imaginative resistance and the Voluntariness condition
My argument up to this point is that the tentative commitment embedded in the presumption of innocence can be characterised as a propositional imagining. But the PoI, as a propositional imagining of the defendant's innocence (in the PWB sense), faces a potential setback in what Tamar Szabó Gendler (2000: 55–56) has dubbed ‘The Puzzle of Imaginative Resistance’: ‘The puzzle of explaining our comparative difficulty in imagining fictional worlds we take to be morally deviant’.
In fictional works, we generally can imagine what an author asks of us, the reader. An author might ask us to imagine ‘By the year 2050, packs of wolves were roaming the towns of England’ 31 and doing so would be fairly straightforward—after all, our capacity to engage in propositional or other imaginings are, as I wrote earlier, generally within our control and thus satisfy the Voluntariness condition. But an author might ask us to imagine claims that we simply cannot or require a ‘very violent’ (Hume, 2000: 14) effort to: that infanticide is morally good (Mahtani, 2012: 415–416; Walton and Tanner, 1994: 37) for example, or that there exists a three-sided circle. This puzzle, as philosophers have recently observed, is not limited to works of fiction but extends to our capacity to imagine more broadly (Gendler, 2007: 149–150). 32 Thus, sometimes when an agent engages in an episode of propositional imagining, she is not only unable to imagine q where q is morally deviant, but also where q is contrary to her ‘aesthetic, semantic, and metaphysical beliefs too’ (Weinberg and Meskin, 2007: 186).
Gendler (2000: 74–74) dissects this puzzle further by observing that the resistance we might feel when imagining certain propositions stems primarily from our unwillingness in so imagining, and secondarily from our difficulty in so imagining. Recognising this, Jonathan Weinberg and Aaron Meskin (2007: 185) divide the puzzle into two types: instances where we are unwilling to imagine q (the ‘puzzle of imaginative refusal’) and instances where we find it difficult to imagine q (the ‘puzzle of imaginative blockage’). I make no attempt here to solve either of these puzzles, for it suffices for my purposes to recognise that these are puzzles of the imagination, and to recognise that these puzzles—where there is a ‘flat-out refusal’ (Weinberg and Meskin, 2007: 185) or ‘full-fledged incapacity’(Weinberg and Meskin, 2007: 185) to imagine—are qualifications to the claim that imaginings are fully subject to voluntary control.
Propositionally imagining the defendant's innocence, I think most would agree, can run into the problem of imaginative resistance but not to the point where the imaginative project becomes an impossibility. When a criminal defendant is prosecuted it may be difficult, or we may be unwilling, to propositionally imagine her as innocent pursuant to the PoI. This may be because the evidence against the defendant is such that—despite falling short of the criminal standard—the fact-finder still finds it difficult to believe in her innocence. In such a case, whilst the mind might resist the idea of the defendant's innocence, that resistance is not complete (in contradistinction to, for example, imagining a three-sided circle or that infanticide is morally good). The Voluntariness condition, therefore, remains satisfied if the PoI involves imagining the defendant's factual innocence.
The fact of imaginative resistance in these hard cases, I think, indicates the importance of conceiving the PoI as an attitude toward the defendant's innocence. To my mind, it is not an objection to the PoI as a propositional imagining that we might from time to time find it difficult to imagine the defendant's innocence; indeed, that such a propositional imagining can be difficult is precisely what I would expect of a rule of law that is supposed to counteract the disposition in the mind of the public (based on the available evidence concerning those who are charged with crimes) that the defendant is guilty (Allen, 1931: 255). It is precisely because the fact-finder is initially inclined to discount the possibility of the defendant's innocence that she finds it difficult to propositionally imagine it, and by instructing her to do so (even when difficult) balances the scales: Imagining can sometimes lead to belief or acceptance, so one may avoid imagining subscribing to a moral perspective that one considers pernicious or reprehensible (or to a false factual claim), for fear of succumbing to it. The danger … may be merely that the person will take it more seriously than she thinks she should, or regard it as less than utterly absurd, as an alternative to be considered … [I]maginings can encourage behaviour in accordance with a point of view one rejects, even if one continues to reject it. It is likely that the more confident a person is of her convictions, the more she will want not to be ‘oriented’ differently, not to be induced by imagining to act contrary to them. (Walton, 2007: 143)
The point is that propositionally imagining the defendant's innocence will face resistance in some cases, but that is exactly what we should expect of the PoI. It is because fact-finders either do not want to, or find it difficult to, consider the flip-side of the coin, and just because we want them to consider the flip-side of the coin, that we instruct them to undertake this imaginative task. The same can be said of other subjects of the PoI—although the police and prosecution may still believe the defendant to be guilty, their propositional imagining of the defendant's innocence has the potential to motivate a more scrutinous and balanced review of the evidence.
Whilst I can accept that this vulnerability to imaginative resistance might not strictly count as credit in favour of the PoI as a propositional imagining, I certainly do not count it as discredit.
Suppositions 33
There is a neat solution to the resistance a fact-finder might feel when she propositionally imagines the defendant to be innocent pursuant to the PoI: instead of conceiving the presumption of innocence as a propositional imagining of innocence, it can be recast as a supposition of innocence. I do not agree that this recasting is necessary.
Like propositional imaginings, suppositions that q can consistently be made alongside a belief that q, and vice versa (Arcangeli, 2014: 609). Magdalena Balcerak-Jackson (2016: 53) writes that because we commonly suppose things (for the sake of argument, for example) that we believe or know not to be true, suppositions are typically ‘compartmentalised’ from beliefs—mirroring how propositional imaginings are ‘quarantined’ from beliefs. Suppositions therefore satisfy the Belief Consistency condition.
Like propositional imaginings, suppositions are under our control. Specifically, it is the relative ease with which suppositions can be engaged that make suppositions a prima facie solution to propositional imaginings. To propositionally imagine q (where q might inherently be unbelievable, or for whatever reason triggers imaginative refusal or blockage) requires, as most philosophers writing on suppositions accept, a greater mental effort than merely supposing q. For example: René Descartes (1996: 50–51) wrote that imagining ‘requires a peculiar effort of mind’; Kendall Walton (1990: 20) has suggested that imagining ‘is doing something with a proposition one has in mind’; and Gendler (2000: 80) has written that ‘imagination requires a sort of participation that mere hypothetical reasoning does not’. By contrast, Weinberg and Meskin (2007: 193) have pointed out that suppositions can easily be made about any propositions, be they ‘patent contradictions [or] ethically repulsive propositions’. Supposing the existence of a round square seems to be less challenging than propositionally imagining a round square, and so supposing the defendant to be innocent has the advantage of dodging the imaginative refusal or blockage that propositionally imagining the same faces.
The comparative ease of supposing difficult propositions, and how a supposition of the defendant's innocence avoids the blockage or refusal that can attend propositional imaginings, might make supposition appear superior to imagination as an explanation of the PoI. Still, I prefer to think of the presumption of innocence as a propositional imagining. Perhaps surprisingly, my reason for this preference is premised exactly on the resistance we feel when we propositionally imagine the defendant's innocence, and the comparative ease of supposing the same. The reason why imaginative resistance obtains in certain instances of imagining q is, as I averted to earlier, because ‘imaginings can encourage behaviour in accordance with a point of view one rejects’ (Walton, 2007: 143). Because imaginings have the potential to orient our mind differently, and because we sometimes do not wish to be ‘induced by imagining to act contrary to [our convictions]’ (Walton, 2007: 143), we feel resistance. All of this demonstrates that propositional imaginings, while they can be difficult, can motivate us to act or re-orient our views in ways that supposing cannot. It is precisely because supposing lacks this motivational power that we feel almost no suppositional resistance in supposing difficult propositions (whereas in a mirror instance we do feel imaginative resistance). If we are genuinely invested in avoiding wrongful convictions, we should strive toward a conception of the PoI which has the cognitive power of forcing the fact-finder to take the innocence of the defendant seriously. A propositional imagining of the defendant's innocence achieves this better than a supposition does.
Conclusion
The idea of a presumption is properly part of the PoI, contrary to the Sceptical argument. One can study the PoI qua presumption by unshackling oneself from two fallacies: first, the thesis that the PoI is a ‘belief in’ the defendant's innocence; and second, that ‘true’ presumptions are evidentiary devices in the form of a modus ponens.
Now free from those fallacies, I have argued that the PoI is a specific type of presumption: a practical presumption that involves a tentative commitment that the defendant is innocent (‘presumably, the defendant is innocent’) and an action policy of proceeding as if the defendant were innocent—all in the name of avoiding the greater harm of wrongful convictions (the other, lesser harm being false acquittals).
The final part of my argument targeted the nature of this ‘tentative commitment’ that the defendant is innocent, bearing in mind that this ‘tentative commitment’ must avoid the twin objections levelled against the Belief Thesis. I have chosen to settle on a representational account of the propositional imagination: the PoI is just a propositional imagining of the defendant's innocence, which has the potential to motivate the action policy of proceeding as if the defendant were innocent (unlike a supposition of the defendant's innocence).
Accordingly, the PoI is a practical presumption, one that involves a propositional imagining of the defendant's innocence (ie, ‘presumably, q’) and an action policy of proceeding as if the defendant were innocent.
Footnotes
Declaration of conflicting interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article
