A. Malleability of Analogy and Accounting for the Element of Choice

Standard accounts of analogical reasoning seem to enter the scene analytically a step too late, as it were. They pay little attention to how the so-called “source” analogue is framed by a court. Judgments of similarity are crucially dependent on how one cognises the case that another is being judged similar in comparison to. To see B as similar to A, one has first to cognise A in a certain way. That acts as the “fixed point” from which one – going by standard accounts – launches analogical reasoning.Footnote ¹³ Framing involves cognising a source analogue as falling within a certain “category or set of categories”.Footnote ¹⁴ The category the source is put into, in turn, is closely tied with whether or not the case gets cognised as an instantiation of some rationale (which is typically wide).Footnote ¹⁵ Once the source analogue is framed as falling under a category, it paves the path for the extension of the rationale to the case at hand, as the examples discussed later in this section shall reveal. This is so because the structure of argument by analogy, in Razian terms, comes to this – if the rationale applies to the source analogue, it ought to apply to the target.Footnote ¹⁶ This makes the initial framing of the analogy crucial.

Standard accounts of analogical reasoning presuppose that framing of the source analogue is an objective exercise which does not admit of an element of choice. As Austin puts it, “it is known (or assumed) before the reasoning that a given something is true” of the source analogue.Footnote ¹⁷ Without this base or fixed point, standard accounts of analogical reasoning can never really get off the ground.Footnote ¹⁸ If the issue of how to frame the source is open for choice on each occasion – not being objectively identifiable at the outset – there is no meaningful way in which one could ever really reason from a source to a target analogue, for it would require the inexplicable prior step of choosing how to frame the source analogue. But to concede that framing of a source analogue is a malleable process involving an element of choice would be to throw a spanner in the workings of the normative model underlying the standard picture.

The gravamen of the charge then comes to this. Since the framing of analogy is a malleable process, involving as it does an element of choice, judgments of similarity in analogical reasoning cannot possibly be normatively constraining. Joseph Raz,Footnote ¹⁹ Neil MacCormickFootnote ²⁰ and Richard PosnerFootnote ²¹ have each argued that analogical reasoning is not normatively constraining, albeit reaching that conclusion through other routes. The route taken here involves arguing that the problem of malleability of analogical reasoning is of a piece with the better appreciated phenomenon of malleability in the context of reasoning with precedent. In both cases, it turns out, the malleability is attributable to the fact that the legal past is not static. It might be in order quickly to rehearse the contours of the problem in the context of reasoning with precedent before seeing how its parallel applies to reasoning by analogy.

It is now trite learning that precedents in the common law do not come with “rules” embedded in them, and that the future court seized of a lis plays an active role in determining what the rule in a past case is.Footnote ²² This is, of course, the most conspicuous reason why searches for an algorithm or test for detecting the ratio of a case have come a cropper.Footnote ²³ But there is another, less obvious factor which equally contributes to the dynamism of the process – a common law decision is capable of being stated and restated in multiple ways. There are two reasons for this. One, is the level of generality in which the holding in a past case can be cast. To use Julius Stone's example – which has since become ubiquitous in discussions on precedent – the holding in Donoghue v Stevenson is consistent with multiple “rules” ranging from the narrow (snails in bottles) to the wide (general duty of care).Footnote ²⁴ The other, is multiple opinions expressed by judges and the opportunity presented to a future court of latching on to any of the opinions and finding the ratio there.Footnote ²⁵ A telling example of this is the entirely novel ratio recently found by the Supreme Court in the century-old decision of Dunlop v New Garage Motors,Footnote ²⁶ by the simple expedient of turning to the opinion of another judge in the case.Footnote ²⁷ Thanks to these features, as Brian Simpson put it, the rules of the common law cannot be engraved on to tablets or counted – as if they were sheep.Footnote ²⁸ In fact, Levi himself – one of the most influential supporters of the standard picture of analogical reasoning – is acutely aware of this phenomenon in case of reasoning with precedent, which prompts him to describe the common law as a “moving” system of classification.Footnote ²⁹ As Grant Lamond points out, without appreciating this malleability of precedent in the common law, one cannot really account for the process of distinguishing.Footnote ³⁰ In a similar vein, it is argued here, the malleability of the common law has implications for analogical reasoning as well. Considerations similar to those found when reasoning with precedent apply mutatis mutandis to analogical reasoning.

It will be argued with illustrations prayed in aid that there is no normatively constraining pathway of extending a past case (the so-called “source”) analogically. And that a prior, independent choice is necessarily involved in what one picks as an analogy and how one frames it. To see this in action, let us consider the movement of the law of frustration, from the celebrated case of Taylor v Caldwell Footnote ³¹ to the early twentieth-century coronation cases. Prior to Taylor v Caldwell, the common law rule – going back all the way to Paradine v Jane (1647) – was that change in circumstances would not discharge contractual obligations. The only exception to the rule was death or perhaps disability in case of contracts of a personal nature.Footnote ³² Taylor v Caldwell was to alter the status quo. The case concerned a contract to hire a hall and gardens adjoining it for a music concert. Before the concert, the hall was destroyed by fire. Blackburn J., relying on Pothier and the Digest (i.e. Roman law) held that “the Music Hall having ceased to exist, without fault of either party, both parties are excused”.Footnote ³³ Taylor had nothing to say on the “teleological notion” of what happens when the object or purpose of the parties becomes impossible.Footnote ³⁴ Blackburn J. himself did not take his decision as upending the old common law rule going back to Paradine v Jane.Footnote ³⁵

Fast forward by a decade to Jackson v Union Marine Insurance where the issue revolved around whether the delay of over half a year caused by the ship having run aground discharged a charter of the ship required for immediate dispatch to San Francisco.Footnote ³⁶ The majority opinion of Bramwell B. found the case analogous to Taylor while the dissenting opinion of Cleasby B. did not find it so. The key to their diverse approaches lies in how they framed Taylor. The dissenting opinion of Cleasby B. framed Taylor as a case of destruction of subject matter. Bramwell B.'s majority opinion framed Taylor by putting it into a category that would lend itself to be extended – namely, one where the teleological notion of the object of the parties came into play. This paved the path for Bramwell B. to see Taylor supported under the broader rationale of frustration of object. Once the rationale of frustration of object is seen to ultimately justify the outcome in Taylor, it was found good enough to justify the outcome in Jackson.

Later, in Krell v Henry,Footnote ³⁷ where the issue was whether a contract was discharged when an event (the coronation) which formed the basis of the parties entering into it did not come to pass, Vaughan Williams C.J. repelled the defendant's argument (which harked back to Taylor) that the contracts could only be discharged for destruction of subject matter – by specifically pointing out that Jackson had “extended” Taylor analogically. It is only after Krell that “frustration of object” became cemented in English law as a ground for discharge of contract. In retrospect, Taylor being invoked as analogy for “frustration of object” might seem to a textbook-bound student to be an inescapable choice. But this would be to ignore the fact that there is no objective or neutral way of framing Taylor, and any way of doing so involves an element of choice. Nor can there be an assumption that the rationale underlying a source analogue must always necessarily be a broad one. English law, for instance, has consciously not extended “impossibility” and “frustration of purpose” cases to cases of “commercial impracticability” where the commercial purpose for which the contract is entered into is radically altered. By contrast, the Uniform Commercial Code (UCC) in the US has extended this line of cases to “commercial impracticability”.Footnote ³⁸ Someone inclined to endorse this might do so upon a framing of Jackson and Krell in a way that supports the rationale of “commercial impracticality”. After all, it is not too far-fetched to extend “frustration of object” to the frustration of the “commercial objective” of the contract. It is, in fact, not even that far-fetched to frame Taylor in such a way as to read the rationale of “common object”, “commercial object” or any other similarly broad rationale which can then be extended to a wide sweep of cases. Accordingly, Posner wearing his judicial hat, found the economic rationale of “risk transfer” underlying the frustration cases such as Taylor – which is then readily extendable to “commercial impracticability” cases.Footnote ³⁹ There is little doubt that the drafters of the UCC too must in their deliberations have reasoned “analogically” from the existing law, but the key to this is to see the choice made by them to frame the past cases in a certain way. And it is as much of a choice which explains why no English court has framed these cases in a way that would enable them to use them as analogies to discharge contracts for “commercial impracticability”.

When allowance is made for the invocation of distant analogiesFootnote ⁴⁰ and the related phenomenon of invoking broadly couched rationales to extend a past decision to the case at hand, one finds that not only is there an ample leeway in how one frames a source analogue, but also in what one chooses to invoke as the source analogue in the first place. The greater the leeway gets, the stronger becomes the presumption against any normatively constraining judgments of similarity from past cases. But the true magnitude of the leeway available to a judge in analogical reasoning does not emerge clearly until one appreciates cases in the opposite direction – where the courts have not extended a case analogically; even in circumstances that would seem “similar” to the greenest matriculand. These often remain hidden in plain sight simply because theory tends not to see them as deliberate choices related to analogical reasoning. But they are; and it is important that they are understood in those terms.

A good example of this phenomenon is found in how Rylands v Fletcher has come to be treated by common law courts, particularly in the post-war period. In the Exchequer Chamber Blackburn J. held that anyone who brings on their land something not naturally there is liable for the damage caused by its escape, even if it was without their fault.Footnote ⁴¹ If there was ever a case which had all the potential of becoming an engine for analogy creation, it was Rylands. In fact, by the early twentieth century, academic commentary had already paved the path for its wide invocation by pitching Rylands as supportive of an idea of “strict liability” for ultra-hazardous activities.Footnote ⁴² But the subsequent history of tort law reveals the lengths to which the courts went to avoid the seemingly easy path of extending Rylands.

In Read v Lyons, a munitions inspector in an ordinance factory suffered personal injuries from an exploding shell during manufacture.Footnote ⁴³ At first instance, Cassels J. found for the plaintiff by extending Rylands to cases regulating “ultra-hazardous” activities. Cassels J. also found that the case was analogous with cases falling under the scienter rule – liability for animals known to be dangerous – which too, he argued, could come within the ultra-hazardous principle of Rylands. On both counts, Cassels J.'s decision involved analogically extending Rylands to cases which did not directly fall within its scope.Footnote ⁴⁴ But the Court of Appeal reversed the decision – which was confirmed by the House of Lords. The House of Lords read the Rylands rule narrowly and found that the case did not fall within it. The primary ground on which they dismissed the claim was that the Rylands rule was not engaged since the injury had occurred on the premises of the defendant and there was no “escape”. But it is interesting to note that, with the “rule” in Rylands not strictly being engaged, the House of Lords did nothing to extend Rylands to the case at hand either. In fact, Lord Macmillan specifically rejected the analogy of liability for personal injuries caused by dangerous animals (the scienter rule) which was found persuasive by Cassels J. in the court below. And Lord Porter noted that, if liability in Rylands which “was itself an extension” were to be extended further analogically, it would lead to “oppressive” results.Footnote ⁴⁵ In order not to extend Rylands analogically, the House of Lords consciously and deliberately framed it in very narrow terms. There were also considered dicta from the bench to the effect that personal injuries would not be actionable at all under Rylands. All this strongly indicates that the House of Lords in Read v Lyons sought to frame Rylands narrowly, as a nuisance case.Footnote ⁴⁶ Since nuisance does not extend to personal injury, Rylands framed as a nuisance case, could not extend to personal injuries either. Later, in Transco plc v Stockport MBC,Footnote ⁴⁷ the House of Lords expressly framed Rylands as a nuisance case and held that it could not therefore extend to cases of personal injury. Now, it is not at all obvious that Rylands is to be necessarily framed in these narrow terms. To consider an example from elsewhere in the common law world, the Supreme Court of India has extended Rylands to cases of liability for personal injuries caused by any “hazardous or inherently dangerous activity”.Footnote ⁴⁸

Another decision that illustrates this consciously narrow framing of Rylands v Fletcher is the decision of the Court of Appeal in Stannard v Gore.Footnote ⁴⁹ The defendant stored a large number of rubber tyres on their premises. The tyres caught fire, through no fault of the defendant, and the fire spread to the neighbour's property where it caused damage. The question was whether Rylands v Fletcher would support imposition of liability in the case. An adherent of the standard picture to whom past cases normatively extend themselves might suppose that, even if one did not subsume this case within the Rylands rule, it would be a tall order indeed to block analogising from it to a liability in the case at hand. There are multiple points of similarity between this case and Rylands which seem to cry out for an extension of Rylands. But here, again, not only did the Court of Appeal find that the case could not be subsumed within Rylands, they did not extend it analogically either. The court read Rylands as requiring that the very thing brought into the land ought to escape – which in this case would be tyres. The fire from burning tyres escaping would not do. One could well argue that this is the equivalent of framing Donoghue v Stevenson as a case confined to harmful fauna in opaque containers.

What the trajectory of the foregoing illustrations establishes is that there is no normatively constraining pathway of extending the past case (the so-called “source”) analogically. And it follows that analogical reasoning cannot be the extending of a past case guided by some normative notion of similarity. A prior, independent choice is necessarily involved in what one picks as an analogy and how one frames it. That choice being a hallmark of analogical reasoning, any plausible theoretical account of analogical reasoning had better explain it. However, there is none forthcoming in standard accounts of analogical reasoning. Indeed, admitting a prior independent choice risks driving a coach and four through the standard accounts of analogical reasoning since it would mean conceding the absence of any normative constraint by the lights of the truth-based model of normativity presupposed by the standard accounts. As we shall find in the objection that follows the present one, this feature of analogical reasoning can be explained by the fact that the direction of fit of analogical reasoning is the reverse of what the standard picture assumes.

B. Rhetoric and Direction of Fit

The argument that analogical reasoning is rhetorical reasoning is a familiar one, but underlying it is a deeper objection about the direction of fit of analogical reasoning, which does not get articulated clearly enough. Let us first consider the familiar “rhetoric” argument.Footnote ⁵⁰ Posner, its foremost proponent, argues that standard accounts of analogical reasoning at best capture how judges “talk”, but that is far from how judges actually “think”.Footnote ⁵¹ As Posner puts it, the past case is merely “information” which the judge is at liberty to use in any way she likes. The judge knows at the outset what she seeks to achieve, and an analogy or example is invoked as a means to persuade the interlocutor to adhere to the speaker's argument. Posner's critique, however, presupposes and points to – if only dimly – a deeper underlying worry about standard accounts of analogical reasoning, of which, the rhetorical point is only the sequitur. That deeper worry is that standard accounts of analogical reasoning have got the direction of fit between the so-called “source” analogue and the “target” analogue all scrambled up. Posner faintly hints at this by invoking Hume in support:Footnote ⁵²

To appreciate the full import of Hume's argument on analogical reasoning, it would be helpful to note the striking parallels here with Hume's own account of morality. Hume argued that ethical judgments are not discoverable by reason and hence are not truth-apt, which is to say, they are not to be assessed in terms of truth or falsehood. Rather, these were expressions of attitudes and emotions – which are mental states not assessable in terms of truth or falsehood.Footnote ⁵⁴ Hume's famous motto was that “reason” is a “slave of the passions”.Footnote ⁵⁵ What Hume does here is propose a “reverse direction of fit” of moral judgments to the one proposed by truth-based models. While truth-based models have a “world-to-mind” direction of fit, the Humean one is based on its reverse: a “mind-to-world” direction of fit. The upshot of this Humean move is that the point of ethical argument is persuasion, and not the invocation of some truth.Footnote ⁵⁶ This is doubly locked in place by Hume's picture of psychology on which world-to-mind judgments (e.g. belief about some truth), are motivationally inert while mind-to-world judgments (attitudes or emotions) are motivationally laden.Footnote ⁵⁷ As we shall see in the following section, persuasion and rhetoric are salient to Hume's account of ethics.

Hume's parallel argument about analogical reasoning is that the direction of fit between the so-called “source” and “target” analogue is the reverse of what the standard accounts of analogical reasoning suppose. The decision to invoke a source analogue is not informed by any normative constraint (it is not prone to truth or falsehood, as Hume puts it) put in place by the source analogue itself, but rather it is the prior, independent decision (preference “founded on taste and imagination”, as Hume puts it) to reach a certain result which informs the choice to frame the source analogue in a certain way. That prior, independent decision having been made, the so-called “source” analogue gets chosen, framed and invoked to justify it. The choice of source analogue is at the service of whatever end the reasoner has in mind. It will be recalled that in contrast to this Humean picture, standard accounts of analogical reasoning presuppose a truth-based model of normativity on which target analogues involve normatively constraining judgments of similarity with past decisions. Although Hume himself does not take the step here – but Posner does – the logical sequitur of Hume's views on the direction of fit of analogical reasoning is that it is a form of rhetorical reasoning where a past case is picked for its value as a rhetorical counter to persuade the interlocutor about the acceptability of a prior, independent decision on the substance of the case. The full implication of these arguments would be, in effect, to reverse the picture of analogical reasoning found on standard accounts. The lesson at the heart of Hume's argument revolving around direction of fit is that the truth-based model of normativity, which is presupposed by standard accounts of analogical reasoning, is a mischaracterisation of analogical reasoning.

Now, Posner makes the familiar rhetorical point only to write off the whole enterprise of analogical reasoning. This is the natural denouement of Posner's rhetoric-based objection. Something that Posner himself never specifically spells out in so many words is his supposition that since what is at play here is rhetorical reasoning, it is unfathomable how it should ever be normatively constraining in any manner. But this sceptical conclusion only follows if one supposes that the truth-based model of normativity provides the only plausible explanation of analogical reasoning. If this were to be the case, any piece of rhetorical reasoning – in which direction of fit runs in the “reverse”, which is to say, from the present case to the past case – would eo ipso be without normative constraints. As we shall see in the following section, this is far from being the case.

A. The Idea of Polycentricity

Polycentricity is an idea due to the polymath Michael Polanyi which he gleaned, to a significant extent, from the workings of the common law.Footnote ⁵⁸ In abbreviated form, the idea comes to this. A polycentric system of decision-making (such as the common law) involves a number of decision makers who go about their task of making decisions by anticipating what is likely to pass muster with the community.Footnote ⁵⁹ And having thus arrived at a decision, then seek to persuade the community of decision-makers by justifying the decision at hand by falling back on previously decided cases. Polycentric decision-making can thus be understood as a two-pronged process. The first prong is epistemological, involving as it does, discovery, while the second prong is rhetorical since it involves justification of the decision.

Lon Fuller borrowed the term “polycentricity” from Polanyi and used it as a rubric to discuss an aspect of common adjudication,Footnote ⁶⁰ but Fuller's point is rather different from Polanyi's.Footnote ⁶¹ Although Fuller's idea of polycentricity (which has little to do with Polanyi's idea of polycentricity) has had some traction among legal scholars, Polanyi's idea remains largely neglected.Footnote ⁶² Here, our attention will be on Polanyi's idea of polycentricity – not Fuller's. In sharp contrast to its fortunes in legal scholarship, Polanyi's idea of polycentricity has left an indelible mark on many other areas of thought including economics (ranging from Friedrich Hayek's idea of the working of spontaneous orders to Elinor Ostrom's idea of the regulation of the commons) and philosophy of science (Thomas Kuhn's notion of scientific paradigms, which is discussed later). One would have to put this underappreciation of Polanyi's idea of polycentricity (hereafter “polycentricity”) in legal scholarship down as somewhat of a missed opportunity, for polycentricity has the potential to deliver illuminating theoretical insights on the workings of the common law. This is all the more so because Polanyi gleaned the idea, in large part, from the working of the common law. Elsewhere, I have sought to sketch out some potential insights polycentricity might hold for theory construction in the common law.Footnote ⁶³ Here, we will see how polycentricity helps us appreciate analogical reasoning in the common law. And, more significantly, we will see how polycentricity holds the key to showing us that even if past cases are not normatively constraining in analogical reasoning – and that there is an inevitable element of choice in selecting and framing analogies – it could nonetheless yield a robustly normative system through other means. This is thanks to the motivational normative model underlying the polycentric account which shall be seen to emerge as an alternative to the truth-based model of normativity.

B. Epistemology of Polycentric Judging

For better appreciating how polycentricity works in the law, it might be useful to begin with one of Polanyi's simpler examples demonstrating how the most elementary of pricing operations in the market display the epistemological (anticipating how the market will react) and rhetorical (pricing it such as to persuade the buyer to purchase) prongs of his polycentric theory. A considerable chunk of Polanyi's oeuvre was dedicated to elucidating the epistemological prong of polycentricity – his most lasting contribution perhaps being the idea of “tacit knowledge”.Footnote ⁶⁴ Knowledge that one can use proficiently, albeit without being able to state it in the form of propositions is “tacit knowledge”. It is found in activities ranging from learning swimming to mastering a language – although there are important differences between knowledge acquisition in the two cases. While the mastering of swimming is a solitary activity, mastering a language involves a communal element. As Michael Oakeshott aptly described it, knowledge of language should not be seen as a mastery of a fixed set of utterances, but rather, as the ability to anticipate what is likely to pass muster with a group of language users.Footnote ⁶⁵ This point is well illustrated by an anecdote related by Brian Simpson – which, as we will find, is of a piece with the legal case. When faced with an objection from a copy editor that the word “inruption” cannot be used in the manuscript since it did not find mention in the Oxford English Dictionary, Simpson replied: “[T]he word was properly formed from Latin roots, ‘in’ and ‘rumpere’ (as the copy editor conceded) and, once it appeared in my book, would at once exist as an English word ... Once in print ‘inruption’ could then be recorded in the Oxford dictionaries, and, as if by magic, become correct English.”Footnote ⁶⁶

The deeper message underlying Simpson's anecdote, as he himself explains it, is that evolution of the English language is akin to the evolution of the common law.Footnote ⁶⁷ At the heart of the evolution of both is tacit knowledge – although Simpson himself does not himself use the idiom of “tacit knowledge”. An adept user of language learns to anticipate which coinage is likely to pass muster with the community of language users. This is the reason the neologism “Covfefe”Footnote ⁶⁸ is unlikely to ever be mot juste for anything sensible, unlike Gustave Flaubert's invention of the latter phrase. Just like an adept language user, a common law judge learns to anticipate what is likely to pass muster with the legal community. Elsewhere, Simpson has described how this works, by sketching the picture of the common law as the collective intuitions of the legal profession.Footnote ⁶⁹ It might be more aptly described as the collective tacit knowledge of the profession. It involves knowing the apt response to the case at hand; the apt response, in turn, is to be understood as an anticipation of what response is likely to pass muster with the community of decision makers. Simpson's argument is that through years of training, the common lawyer tends to internalise the gaze of the professional group and its sensibility of what is the apt response in a given case.Footnote ⁷⁰ By the time they are elevated to the bench – typically drawn from senior members of the bar – they tend to become experts at anticipating what is likely to pass muster with the legal community.Footnote ⁷¹ Thomas Kuhn argued that scientific research too works in the same way with the “tacit knowledge” in this case being held up by the collective gaze of the peer group of researchers.Footnote ⁷²

The legal practitioner's skill lies in anticipating – through a similar process of tacit knowledge – what approach a judge is likely to take in a given case.Footnote ⁷³ It is this simple insight which seems to be at the core of Oliver Wendell Holmes Jr.'s notion of the law as a “prophecy”,Footnote ⁷⁴ which tends – rather unfairly, one might add – to be given short shrift by legal philosophers.Footnote ⁷⁵ What Holmes suggests makes sense of why legal opinions are typically couched in the language, not of what the law “is”, but rather in terms of how the court is “likely” view a matter. This also ties in with Holmes’ oft-quoted but poorly appreciated idea of “experience”, which was very similar to the idea of tacit knowledge.Footnote ⁷⁶ The word “experience” has this inherent ambiguity since the Greek emperikos (experience), to which it traces its provenance, shares the same etymological root as empeiros which means “experienced” in the sense of being “skilled”. This image of the common law is one that fits in quite nicely with Polanyi's polycentric picture. Historically, the idea of “artificial reason” going back to Coke – deriving from an artifex or craftsman – lends itself to be understood in these terms.Footnote ⁷⁷

Depictions of the common law consistent with this idea of tacit knowledge can also be found in the writings of Friedrich Hayek,Footnote ⁷⁸ John Baker,Footnote ⁷⁹ Anthony Kronman,Footnote ⁸⁰ John Bell,Footnote ⁸¹ Stanley FishFootnote ⁸² and, more recently, in the work of Hanoch Dagan.Footnote ⁸³ Although Karl LlewellynFootnote ⁸⁴ neither systematically developed the idea nor placed it in a broader philosophical account, there are elements in his account that accord with this idea of tacit knowledge. The idea is also compatible with those who tend to view common law judging as involving a form of “aesthetic sense”,Footnote ⁸⁵ practical wisdom or techne Footnote ⁸⁶ – just so long as one bears in mind the communal element involved in its exercise, which is not necessarily present in all examples of practical wisdom or phronesis.Footnote ⁸⁷ As we shall see later, the communal nature of the tacit knowledge is the all-important element in the model of normativity underlying polycentric systems. Relying upon such tacit knowledge, one might anticipate – although not anywhere near as skilfully as a practitioner, let alone a judge – that English courts are presently unlikely to extend the “strict liability” regime inaugurated by Rylands. Seen thus, the judicial choice underlying framing Rylands narrowly and not extending it in a succession of cases becomes transparent. No amount of staring at cases or plotting out the lines of similarity emanating from them can replace this skill in anticipation.

Adherents of standard accounts of analogical reasoning like Weinreb, who emphasise the involvement of skill in the process, dimly touch upon something vital, but stop short of coherently articulating it, given the model of normativity their accounts are enmeshed with. On these accounts, legal skill ends up as little more than a parlour trick as its principal task is to connect past cases to present ones through likeness and similarity.Footnote ⁸⁸ Seen thus, it becomes akin to a “matching of colours”, to use Benjamin Cardozo's memorable phrase.Footnote ⁸⁹ But this is to mischaracterise what is understood as acquisition of skill in the law. The common law has a self-conscious conception of skill (artificial reason) going back all the way to Coke. This is best understood as sensibility which guides a judge to what is the apt substantive decision in the case with reference to what is likely to pass muster with the group.

This may be as good a place as any to enter a caveat about the nature of the tacit knowledge being discussed here. Three factors moderating the efficacy of tacit knowledge may be noted. First, as Kuhn argued, tacit knowledge explains how “normal science” works, but there are also periodic ruptures brought about by “scientific revolutions” which tacit knowledge cannot anticipate.Footnote ⁹⁰ Likewise, as Dagan points out, there are going to be paradigm altering cases which cannot be readily anticipated through tacit knowledge.Footnote ⁹¹ Arguably, the decision of the Supreme Court in the recent “prorogation” case could be thought to be of such a kind.Footnote ⁹² Such cases excepted, it would not be rash to say that tacit knowledge can be counted upon to do its job reliably – which is to say, more than “eight times out of ten”.Footnote ⁹³ Secondly, tacit knowledge is better at anticipating what a bench will hold rather than what an individual judge will. There will be cases where an individual judge, despite being mindful of the result dictated by tacit knowledge, might choose to chart a different course – as it is typically the case in dissenting opinions. However, it is important to note that these are viewed by the legal community as deviations which are policed by force of the collective professional gaze that holds up tacit knowledge. Commonly, this takes the form of thwarting the “reformist” agendas by reinforcing the “orthodoxy” – as Lord Mansfield repeatedly faced.Footnote ⁹⁴ In rare cases, it could also take the extreme form of the kind of ostracism Lord Atkin had to endure after his dissent in Liversidge v Anderson.Footnote ⁹⁵ Thirdly, from the practitioner's point of view, what tacit knowledge is likely to reveal more reliably is the result or relief in the case rather than the route through which it is reached. A good number of Lord Denning's ostensibly radical decisions reached perfectly predictable, even anodyne, results in the case at hand – on which he was in agreement with the rest of the bench – albeit by reasoning to it through the use of a principle thought too broad or wide by the others on the bench. Lloyds Bank v Bundy Footnote ⁹⁶ (where he introduced the doctrine of “inequality of bargaining power”) and Butler Machine Tools v Ex-Cell-O Corporation Footnote ⁹⁷ (where he proposed abolishing orthodox doctrines of “offer” and “acceptance” as a rubric to understand contract formation) are good examples of this.

C. Rhetoric and Justification

Once the judge anticipates or knows (guided by tacit knowledge) the apt response in the case at hand, we move to the next step of the process – which is that of persuasion.Footnote ⁹⁸ Here, the judge falls back on the stock of past responses and seeks to justify the decision in the case at hand by showing how it is similar to a previous case. Past cases then, are used as rhetorical counters which are meant to persuade the interlocutors about the decision in the present case. Although Hume himself never explicitly drew out this connection, it follows from Hume's account of the common law – touched upon briefly in Section 2 – that “analogical reasoning” is a form of rhetorical reasoning. To help us understand how a rhetorical account of law and persuasion follows from Hume's starting premises, we would do well to cast a glance at the nineteenth century Scottish philosopher, Joseph Campbell's Philosophy of Rhetoric in which he developed a theory of rhetoric based on a Humean theory on moral values.Footnote ⁹⁹ At the heart of persuasion and rhetoric, for Campbell, was the Humean idea of attraction and association among impressions. The central function of rhetoric was the transfer of force from one impression to another by relying on resemblance and contiguity.Footnote ¹⁰⁰ Persuasion would be more effective if built upon relations of resemblance and similarity with already accepted opinions.Footnote ¹⁰¹ Perelman has very similar things to say on the persuasive value of resemblance and similarity, albeit without basing it on a Humean account of practical reasoning.Footnote ¹⁰²

Seen thus, analogical reasoning in the common law is a form of rhetorical reasoning in which past decisions are used as rhetorical counters in a discursive enterprise to persuade the legal community. Once the judge knows through tacit knowledge what the apt response in a given case is, the next step is to “justify” the decision in these rhetorical terms by using past cases which will give traction to its persuasive value. In the context of analogical reasoning, this means extending a past decision. As discussed earlier, this involves framing the case in a certain way so it may provide rhetorical support to the conclusion reached in the case at hand. This account, therefore, makes sense of the leeway found available to the judge in framing an analogue. On this picture, a judge approaches past cases (putative analogues), having already made a choice about how to frame them – whether in wide or narrow terms. This makes analogical reasoning a part of the “justification”, rather than the process of “discovery”. This is something sceptics of analogical reasoning like Posner harp on, as if it could be nothing but a fatal flaw. On the alternative account being sketched here, it is seen transformed as a part of the polycentric account.

The term “rhetoric” – which is depicted as an important element in analogical reasoning in this alternative account – is likely to set off fog horns up and down the ranks of modern legal theorists. For some sceptics, the worst thing they could possibly say about analogical reasoning is that it is nothing more than rhetorical reasoning, tout court. And that is supposed to be a knock down argument against analogical reasoning. As we have seen earlier, Posner makes this argument against analogical reasoning. Lest the modern pejorative association of the term colour its proper appreciation, it is important to bear in mind that rhetoric was once a crucial constituent of the philosophers’ repertoire. Isocrates, a contemporary of Socrates – and author of Antidosis and Aeropagiticus – made persuasion and rhetoric the mainstays of his philosophy.Footnote ¹⁰³ It might be noted parenthetically that the idea of scattered rationality plays an important part in Polanyi's idea of polycentricity. And Polanyi's inspiration for the idea of scattered rationality was John Milton's idea of the “scattered” statue of truth in Aeropagitica, named after Isocrates’ work.Footnote ¹⁰⁴ As we shall argue later in this section, the idea of scattered rationality supports rhetoric and persuasion. Aristotle's deep interest in rhetoric should be seen against this historical backdrop, as should the importance given to it by Roman philosophers, Cicero and Quintilian – all of which makes Plato's polemic against rhetoric in his dialogues, Gorgias and Phaedrus, somewhat of an outlier in the ancient philosophical world. This is, no doubt, a far cry from modern philosophical trends where “rhetoric” itself has become a pejorative term. Albert Jonsen and Stephen Toulmin have retraced the trajectory of that downfall by showing us how rhetoric continued to be regarded as a respectable philosophical discipline until the Enlightenment and Blaise Pascal inaugurated much of the negativity now associated with the term.Footnote ¹⁰⁵

Returning to the greatest philosophical backer of the idea, Aristotle defined “rhetoric” as discovery of the available means of persuasion.Footnote ¹⁰⁶ For Aristotle, rhetorical reasoning pertains to reasoning from endoxa or accepted opinions.Footnote ¹⁰⁷ These formed the basis for topoi (topics) or common places (also the title of Aristotle's work Topics), which were widely accepted points that a speaker reasoned from.Footnote ¹⁰⁸ The structure of such argument involved showing how the case at hand was similar to a past case and hence ought to be accepted.Footnote ¹⁰⁹ This form of argument formed a crucial component in Greek and Roman legal cultures.Footnote ¹¹⁰ And it forms an important part of the common law tradition.

As Theodore Viehweg notes, it is no accident that it is the English term for Aristotle's endoxa, namely “opinion”, which is “still used to characterize judicial decisions”.Footnote ¹¹¹ Rhetorical learning was once an important part of the common lawyer's training and, until the birth of modern university legal education, late in the nineteenth century, law was regarded a subdivision of rhetoric.Footnote ¹¹² Rolle's Abridgment, a long-standing mainstay of common law literature was nothing more than Sir Henry Rolle's topoi or common place book.Footnote ¹¹³ The claim that the common law is best understood in terms of topoi and rhetoric is hardly a novel one. It provides the backdrop for Viehweg's magisterial Topics and Law; and ties in with Hans-Goerg Gadamer's claim that common law reasoning is rhetorical.Footnote ¹¹⁴ It was also Julius Stone's view that common law reasoning is a form of rhetorical reasoning, with past cases taking the place of topoi.Footnote ¹¹⁵ In a similar vein, Fish postulates “persuasion” as the “goal” of judging.Footnote ¹¹⁶ The novel theoretical move proposed here would be to place this rhetorical understanding of the common law within a sound model of normativity which has robust normative constraints in place. We now turn to see what this model of normativity looks like.

D. Motivational Model of Normativity

Hume opens An Enquiry Concerning Human Understanding by adumbrating two “manners” or “species” of philosophical enterprise.Footnote ¹¹⁷ The one model envisages the task of moral philosophy as that of informing interlocutors. The other model envisages the task as that of motivating them. The one model takes moral philosophy as a matter of “speculation”, the chief task of which is to expound the “foundations of morals” which it does, by discovery of standards. The other takes us as “active beings” motivated by “taste and sentiment” and hence takes philosophy as aiming for “moulding the heart and affections”. On the former model, normativity or bindingness is a matter of our judgments conforming to some objective or mind-independent standards. As such, our response to or grasp of them is truth-apt, which is to say, capable of being true or false in relation to these standards. By contrast, on the latter model, bindingness or normativity of morality is nothing but the motivational pull exerted by a moral judgment.Footnote ¹¹⁸ On this model, the interlocutor and motivating her take centrality. These models are the truth-based and motivational models of normativity, respectively. Hume himself was sympathetic to the motivational model of normativity. It should be readily apparent why this model of normativity naturally ties in with rhetoric – the principal aim of which is to seek motivational traction with the interlocutor. As Pierre Hadot shows us, the motivational model of normativity was crucial to “ancient philosophy” since much of it was about “transforming” the interlocutor.Footnote ¹¹⁹ Aristotle's Rhetoric works on this model, as did Campbell's Humean theory of rhetoric.

Translated to the case of analogical reasoning, the two models of normativity come to this. The truth-based model has it that there is an objectively correct notion of similarity or likeness which the decision maker is responsive to in extending past cases. And it has it that invocations of analogy are truth-apt in relation to those objective notions of similarity or likeness. Standard accounts of analogical reasoning presuppose the truth-based model of normativity. These objective notions of similarity or likeness are the normative constraints on this model. By contrast, the polycentric account of common law reasoning presupposes a motivational model of normativity. On this model of normativity, decision-making seeks to be persuasive, which is to say, to have motivational traction in the practical reasoning of the interlocutor. On the polycentric model, a decision maker seeks to anticipate what is likely to be persuasive to the community. This leads to a form of scattered or “collective rationality”.Footnote ¹²⁰ The polycentric decision-making is supposed to harness this collective rationality. What is most likely to be acceptable to the group becomes, on this model, the most apt and hence, normatively recommended course to follow. On the polycentric model, it is the anticipation of what is likely to pass muster with the group which constitutes the normatively correct response to the case (these are the normative constraints on this model). And the invocation of a past case on this model serves as a rhetorical counter meant to persuade the interlocutor.

On the motivational model of normativity then, there are no normative constraints on extending the cases, which arise from the cases themselves. There are nevertheless genuine normative constraints on the decision maker. The decision maker is kept in check by tacit knowledge or sensibility of what is likely to pass muster and by perceived limits of what the legal community is likely to find persuasive. Our “ways of going on” are sufficient to hold normativity firmly in orbit without the aid of any objectively given normative constraints in past cases.Footnote ¹²¹ In fact, given the standard picture of psychology going back to Hume – on which it is not “truths” that motivate, at any rate, but rather our attitudes and dispositionsFootnote ¹²² – nothing could be more stable and conducive to a normative account than a lifetime of internalisation which goes into anticipating what is likely to pass muster with the group.

One would only be setting up an Aunt Sally in supposing that, if something akin to the motivational model were to hold, we would all be plunged into a Jerome Frank-ish nihilistic legal universe; one where “anything goes”, hunching reigns supreme and breakfasts dictate legal outcomes. Nothing of that sort need happen at all. Far from it, as Thomas Kuhn's work shows us, even in scientific research – ostensibly the gold standard for objectivity and truth – the truth-based model of normativity does not quite hold.Footnote ¹²³ Even in scientific research, normative constraints are generated by a process of internalisation through membership of a peer group and held in place by a sense of anticipation of what will pass muster among that group. In fact, Friedrich Nietzsche attributed the persistence of morals entirely to internalisation.Footnote ¹²⁴

Since the standard picture of analogical reasoning and the truth-based model of normativity underlying it have been enormously influential, the legal theorist's intuitive response here might be to suppose that underlying the motivational model must be the truth-based model of normativity after all, so that, properly understood, the motivational model is no genuine model of normativity at all.Footnote ¹²⁵ There are two strands to this response which we would do well to parse out.

One strand supposes that underlying the motivational model is some normatively correct – where “correct” is understood by the lights of the truth-based model – way of reasoning analogically from past cases which is what is being tracked by this tacit knowledge with the result that the motivational model ends up elliptically referring back to the truth-based model of normativity. It is, however, of some significance to recognise that seen from a metaethically aseptic or neutral viewpoint this would have us skewing the problematic of normativity in a manner that privileges one conception of normativity (namely the truth-based model) and in limine rules out its rival conception (namely the motivational model).Footnote ¹²⁶ Part of the trouble comes from the very term “normativity”. The term “norma” from which “normativity” derives comes from the Latin term for builder's square. From a metaethically aseptic viewpoint, an account of “normativity” must explain how an activity guides and places demands on its participants. It is an unfortunate etymological concatenation that the term “norm” tends now to be used interchangeably with “rule” or “standards” – which is something that has only served to skew the problematic of normativity. For in detecting “normativity” one is then primed to look for “norms” or “rules” underlying any practice. Doing this, however, would be unwittingly to load the dice in favour of the truth-based model of normativity. From a metaethically aseptic viewpoint, it must be acknowledged that there could be more than one plausible second-order account of the “normative” demands an activity places on its participants. One is the truth-based model and the other is the motivational model. On the motivational model, it is the anticipation by the group of what is likely to pass muster which constitutes the normatively appropriate response to the case. Hence according to the adherent of the motivational model of normativity, it would not be proper to dismiss her account of anticipation as a merely “non-normative” or “predictive”.Footnote ¹²⁷ To illustrate the difference in the operation of the two models of normativity, it would be helpful to consider the case of natural language, which is a normative practice. The truth-based and motivational models of normativity seek to provide rival, second-order accounts of this normative practice. The truth-based model of normativity offers what is undeniably the most commonsensical account of the practice. On the truth-based model there are “rules” or “standards” of language which impinge on users of language – to borrow Simon Blackburn's expressionFootnote ¹²⁸ – who in turn respond to the “rules” or “standards”. Uses of language are truth-apt in relation to these rules or standards. The motivational model of normativity offers a rival, alternative account of this practice. On this model, the community's anticipation of what linguistic response will pass muster with language users constitutes the normatively appropriate response. It would be to misunderstand the distinct method of operation of this model if anticipation of what is likely to pass muster with the group is thought of as elliptically referring back to the correct rules or standards. On this model, uses of language are not truth-apt in relation to some rules or standards – not even in “easy” cases where controversy does not ordinarily break out, such as elementary spelling. According to the motivational model, in easy cases no less than in complex and potentially controversial cases (e.g. minting neologisms such as Simpson's “inruption”), normativity is a function of anticipating what is likely to pass muster with the group. The only difference between the easy and complex cases is here the stability of the shared responses. In easy cases, the shared responses are very stable and it is relatively easier to anticipate, even by novices; hence viewed ab extra they tend to have a standard-like feel to them. Anticipation in more complex cases requires a stronger grasp, even mastery, of the practice and tends to have more of a skill-like feel to it. On this point, there are striking parallels with Wittgenstein's rule-following arguments in which he had argued that grasping a “rule” should be seen, not as extending something inherent in the rule itself, “as if they had rails laid to infinity”, but rather, as anticipating the shared responses of the community of rule-followers.Footnote ¹²⁹

The second strand of the legal theorist's intuitive response might be to suppose that “motivational” normativity could never be genuine normativity since motivation cannot really come apart from “rationality” and that one cannot motivate interlocutors into doing “irrational” things – not systematically at any rate. Hence, what must truly motivate interlocutors, the objection goes, is genuinely “rational” requirements which should remain unaccounted on the motivational model. Notice how this response too supposes the truth-based model of normativity and tacks it onto “rationality”. The supposition here is that there are such things as “standards” of rationality which impinge on agents, which they ought to be responsive to. But this again would be to load the dice in favour the truth-based model of normativity. According to the “standard picture of psychology” discussed earlier, widely endorsed by philosophers across the metaethical spectrum, states of mind are divided into cognitive states such as beliefs (motivationally inert) and conative states such as attitudes (motivationally active). What are classified here as “rational” requirements or standards would fall on the “cognitive” side of the table which means that they would be motivationally inert. Any motivation would necessarily involve having to affect the interlocutor's conative states (e.g. attitudes) – which is the central point of rhetoric as Campbell put it. Hence, the supporters of the motivational model will counterclaim that it is their model alone which explains motivation, not the truth-based model. If this response leaves the legal theorist with the residual unease that the motivational model lacks “normative standards” external to the communal or practice-generated consensus – which is the fulcrum of the motivational model – it may be pointed out that this is the fundamental point on which the motivational model differs from truth-based model. On the motivational model, there is no standard or requirement beyond the practice which normatively regulates moves within it. The “ways of going on” are the source of normativity. Therefore, this point again only serves to illustrate the difference in the working of the two models rather than show that the motivational model is wrong.

All said and done, this paper does not seek to argue that the truth-based model of normativity is wrong or that the motivational model is right. To be sure, the onus of establishing that metaethically would be a heavy one indeed and cannot even begin to be discharged within the interstices of a paper on analogical reasoning in law. Rather, this paper simply rests on a plea to recognise that, from a metaethically aseptic viewpoint, there are two plausible models of understanding normativity; one is the truth-based one while the other is the motivational model. And it rests on a plea, that the motivational model of normativity should not in limine be ruled out, as it tends to be. As such, it makes the case for bringing into reckoning a philosophical account which tends to be neglected by legal theorists.

The truth-based model of normativity tends to have such a strong hold on legal theorists that it is often easy to overlook the availability of an alternative. To this must be added the negative image of rhetoric which gives scholars further cause to give a wide berth to the motivational model of normativity. As a consequence, a philosophically robust alternative to the truth-based model of normativity in common law reasoning has not been forthcoming. This is where the polycentric account based on a motivational model of normativity holds such promise. It provides a theoretical alternative to the truth-based model and has the potential to co-opt theoretical positions about the common law based on tacit knowledge (e.g. those of Llewellyn and Dagan) and persuasion and rhetoric (e.g. those of Viehweg and Stone) which are opposed to the truth-based account, within the fold of a unified theoretical rubric spanning both epistemology and justification. The value of this unified theoretical rubric that combines tacit knowledge (anticipatory prong) and rhetoric and persuasion (justificatory prong) cannot be overstated. Even in the accounts of those like Fish who endorse both tacit knowledge and rhetoric, the two operate in separate silos rather than coming together in an integrated whole as they do in the polycentric account. Neither (1) the anticipatory prong pertaining to tacit knowledge, nor (2) the justificatory prong pertaining to persuasion, taken in isolation, is capable of yielding normativity which is a plausible alternative to the truth-based model. Taken in isolation, an account which relies on anticipation alone or persuasion alone runs the risk of coming across as an unstable mishmash of elements mimicking the truth-based model of normativity. Thus, the act of anticipation could be charged with elliptically referring back to true standards which agents seek to respond to – much like the intuitive response of the legal theorist as we have seen previously. It would need more to anchor normativity in an alternative account. Taken alone, the justificatory prong pertaining to persuasion alone runs into a problem of a different kind. It would not be immediately obvious what are the normative constraints on this account. Without the assurance of normative constraints, any persuasion-based account carries the risk of morphing into an “anything goes” or a pejorative variety of rhetoric. An illustration of this point is the criticism directed against a particular strand of Humean accounts of ethics namely, non-cognitivist accounts such as those of A.J. Ayer, Charles Stevenson, Simon Blackburn and Allan Gibbard. And among the modern non-cognitivists’ attempts to parry these criticisms is one involving falling back on something akin to normative constraints generated by anticipation of what will pass muster with the community, of the sort discussed here.Footnote ¹³⁰ It could, in fact, be argued that Hume's account of “artificial justice” combines the two prongs although they are not brought together as clearly as they are in Polanyi's account.Footnote ¹³¹ The key contribution of Polanyi's account of polycentricity lies in unifying the two prongs of anticipation and rhetoric under the rubric of a unified account where the two dovetail into each other to yield genuine normativity without either prong rendering the other redundant. The basic building block of normativity in this model is what will be accepted by the group. The agent seeks to have her piece of reasoning pass muster with the group. The anticipatory prong is what lets the agent know what will pass muster with the community. And the justificatory prong is the action that actualises it. Both prongs look in the same direction, as it were – which is to say, towards what will pass muster with the group. To return to the pricing example considered earlier, it involves both anticipating how a market will react at different price points and actually offering a product for sale in the market with certain specifications at a certain price point. Neither prong is otiose or redundant. In the legal case, the anticipatory prong which gives a reasoner the sense of what is likely to be acceptable to the group combines with the justificatory prong involving articulation of reasons to form what is called “legal reasoning”.