RUTH G. MILLIKAN'S CONVENTIONALISM AND LAW

Abstract

Conventionalism once seemed an attractive way to justify the viability of the positivistic social thesis. Subsequent criticism, however, has significantly lessened its attractiveness. This paper attempts to revive jurisprudential interest in conventionalism by claiming that positivists would profit more from the conventionalism of Ruth G. Millikan than that of David Lewis.

Three arguments are proffered to support this contention. First, Millikan's conventionalism is not vulnerable to the major criticism leveled at conventionalism, viz its compliance-dependence (i.e., the main reason to follow a convention is that other social actors do so), as this is not its defining feature. Second, Millikanian conventionalism retains conventionalism's ability to explain how law emerges from social practices while avoiding the main disadvantage of Lewisian conventionalism, viz its inability to explain the normativity and contestability of law. Third, Millikan's conventionalism can more effectively repel Dworkin's and Greenberg's assaults on legal positivism than its Lewisian counterpart.

To the memory of Maurice O'Brien

I. INTRODUCTION

Conventionalism once seemed a philosophically attractive way to justify the viability of the positivistic social thesis. This was when David Lewis's concept of “coordination conventions”¹ was drawn upon by H.L.A. Hart in the Postscript to his Concept of Law to discuss the nature of the rule of recognition (RoR).² Subsequent criticism,³ however, has lessened its attractiveness to the point where it sometimes appears to be a minority position in the positivistic camp.

This paper attempts to revive jurisprudential interest in conventionalism by arguing that legal positivists turned to the right stable but bet on the wrong horse. Instead of the heavily criticized Lewisian conventionalism,⁴ which has always been their first choice,⁵ positivists interested in conventionalism should have chosen a radically non-Lewisian understanding of conventions, viz that of Ruth Garret Millikan, which has so far had minimal impact on jurisprudence. A major recent publication on legal conventionalism includes only one reference to Millikan's work, compared with more than 150 references to Lewis's.⁶ My main contention is that Millikan's conventionalism has much greater explanatory power when applied to law than does its Lewisian counterpart.⁷

This obviously raises the question of why yet another conventionalism needs to be added to the heap. There are several reasons, explained in more detail below but worth briefly summarizing here.

First, the key reason for finding that Lewisian conventionalism failed to meet jurisprudential expectations was its compliance-dependence.⁸ As the critics correctly pointed out, a conventional rule is not obeyed simply because others obey it. The key differences between Lewis's and Millikan's approaches are that Millikan does not claim that universal, or even general, adherence to a convention is a necessary precondition for its existence, and her focus is on its ability to perform its “proper function.” In other words, in Lewis, conventions work because they are generally obeyed, whereas in Millikan, they are generally obeyed because they work. Unburdened by compliance-dependence, Millikanian conventionalism is worthy of serious consideration on the part of legal philosophers, as it provides an attractive philosophical framework for jurisprudence.

Second, without compliance-dependence and with its focus on proper function, the Millikanian conventionalism retains the advantages of Lewisian conventionalism while avoiding its disadvantages. The initial interest in Lewisian conventionalism lay in its ability to explain arbitrary, contingent features of legal rules and how they could emerge from a social practice. However, its inability to explain several crucial characteristics of those rules, including their contestability and normativity, exposed it to criticism. Millikan acknowledges a (limited) arbitrariness of conventions and their social pedigree, while providing a solution to both the contestability and normativity conundrums. As is shown below, the former is solved due to the lack of compliance-dependence, the latter by treating the proper function as the source from which the normativity of conventions arises.

Third, Millikan's novel arguments concerning the contestability and normativity of conventions effectively refute two anti-positivist challenges, viz those of Dworkin⁹ and Greenberg.¹⁰ It can more effectively repel the former, as it proposes a congenial explanation of theoretical disagreements, and can accommodate the latter, because it treats a conventional practice's proper function as an internal, value-based (but not necessarily moral) factor that transforms social facts into legal content.

No argument in this paper should be construed as claiming that the nonconventionalist versions of positivism cannot marshal convincing arguments against anti-positivistic criticism (they can) or that there are no other effective ways of characterizing the practice of recognition in a nonconventionalist way (there are). The only thing I want to show is that Millikan's conventionalism can help advance new arguments to counter anti-positivist criticism and can propose a new perspective—compatible with that of Hart—on the recognitional practice.

This paper is structured as follows. First, Millikan's approach to conventions as reproduced behavioral patterns is presented. The reason these patterns are reproduced is their proven effectiveness in delivering results that are advantageous to the cooperating partners (i.e., in performing their “proper function”), and thereby provide a survival value both to the individuals who use the conventions and to the groups to which these individuals belong.¹¹

The second section of the paper demonstrates that Millikan's approach has more explanatory power when applied to legal conventions than does that of Lewis. This section focuses on the five points of comparison between the Millikanian and Lewisian approaches to conventions and specifically addresses: (1) whether they are practice-based or rule-based; (2) whether and to what extent they are arbitrary; (3) the required level of adherence thereto in a given community and whether and to what extent they are compliance-dependent; (4) whether their primary function is to ensure coordination; and (5) whether they are normative. Within each of these points, the advantages of Millikanian conventionalism in elucidating the nature of legal conventions are shown.

II. MILLIKAN'S CONVENTIONALISM: AN OVERVIEW

Philosophers of law generally regard Millikan's contribution to the theory of conventions as being confined to two papers in which she criticizes Lewis's approach.¹² However, the broad philosophical background of her work has to be considered before her contribution to the analysis of conventions can be fully appreciated.¹³

Both Lewis and Millikan came to analyze conventions through their initial interest in language. Lewis was attempting to solve Quine's conundrum about how language could emerge without agreement, as agreement requires some form of language.¹⁴ Millikan was devising an original, naturalistic theory of language and mind—“biosemantics”—grounded in evolutionary biology.¹⁵ This theory explains the crucial notions of the philosophy of mind and language, including such notions as beliefs, intentions, mental representations, and reasons, by reference to their ultimate biological functions and their dependence on biological and cultural evolution.¹⁶

According to Millikan, language is a survival tool that helps navigate a complex world by facilitating the collection, storage, and transfer of information about it (descriptive language), and the articulation of ideas on how to improve it (normative language). Millikan points out that “language is merely a very large set of extant (token) precedents of usage.”¹⁷ Linguistic signs proliferate by being copied and forming “lineages.”¹⁸

The notions of proper function and survival value are crucial to Millikan's theory. Proper function is defined by Millikan as

a thing's own function, so called because it is this function that it has been selected or retained for during the development of the species, the individual or the culture, hence the function, the serving of which by its ancestors accounts for its existence.¹⁹

Because behaviors copied by humans bring about desired effects (i.e., they perform their proper function), thereby making them better off, they have a survival value, i.e., they enhance the organism's and/or group's fitness. The word “function” in the term “proper function” denotes the causal role attributed to a particular item or behavioral pattern and consists in the result(s) it is likely to produce. Survival value is also a function but is better understood as a global fitness-enhancing factor. Several behavioral patterns, together constituting a system of conventions, can each have a causal role. However, whether the system as a whole has a survival value depends on the combined effect of the causal roles of its individual behavioral patterns on the fitness of the organism or group.²⁰

Proper function and survival value together constitute the mechanism underlying the proliferation of language: descriptive language (when true) makes humans better off, as it allows them to form accurate beliefs about the world and base their actions thereon, thereby shortening the odds of finding and using affordances. An effective cultural convention operating in a community may bring about a survival value for the community and its members. Improved coordination (e.g., traffic rules), conflict avoidance (e.g., linguistic conventions that enable mediation), and controlled mechanisms for expressing emotions (e.g., mourning conventions) are all adaptations whose adoption improves the chances of survival for the community and its members.²¹

Therefore, it is of crucial importance for society to have a descriptive language and to ensure that it accurately mirrors the world. Directive language is also useful, as it is a tool for adjusting the world to the state indicated by the directives. When directive language is effective in shaping the world, it realizes its proper function and has a survival value for its users (both speakers and hearers), and the instances of using it are further copied.

The notion of survival value is emphatically not to be understood in the narrow sense of the biological survival of a single individual. The proper use of linguistic conventions is an adaptation to the (social) environment that enhances the fitness of an individual (direct fitness), the fitness of other individuals in the group (indirect fitness), and the fitness of the group to which those individuals belong (group fitness).²²

Anti-psychologism is an important feature of Millikan's theory. She argues that human interactions, and especially linguistic interactions, do not require a great deal of knowledge about the minds of others, including their mental state (beliefs, intentions, etc.). Her anti-psychologism is conspicuous in her devastating criticism of H.P. Grice²³ and in her criticism of “meaning rationalism.”²⁴ In her conviction that Grice was wrong in his account of language, and especially in his belief in the significance of intentions in human communication, Millikan differs radically from Lewis.²⁵

Millikan significantly reduces the cognitive burden imposed on social actors by other theories of mind and language. This reduction primarily concerns the scope of knowledge about others’ internal states (e.g., intentions and preferences). While Lewis believed that knowing what others thought was a prerequisite for coordinating our behavior, Millikan believes that it is enough that each of us is thinking about the same thing.²⁶ This difference is crucial for the way Millikan defines conventions, which—in contrast to Lewis—does not include any reference to people's preferences.

Millikan's anti-psychologism is also crucial to understanding her dissension with those philosophers whose theories of conventions are predicated on mental agreement between social actors, and on shared agency and plans, e.g., Gilbert's theory of conventions by fiat and Bratman's planning theory.²⁷ All these approaches rely on an intricate theory of mind and require telepathic social actors. Millikan not only denies the necessity of such mindreading, but claims that—being cognitively too demanding—it would actually hinder coordination.²⁸ For Millikan, social actors need to be able to read the signs around them, not each other's minds, in order to coordinate their behavior.²⁹

Millikan's anti-psychologism, based on the assumption that no meeting of the minds is necessary for humans to cooperate, will be of use in the further sections of this paper, when the fundamental jurisprudential challenge to Lewisian conventionalism, viz its inability to accommodate the immanent contestability of legal practice, is addressed.³⁰

The need to reduce the cognitive burden also leads Millikan to support the thesis that following conventions involves several kinds of nontransparency, understood as incomplete knowledge about the nature of conventions. Nontransparency concerns the arbitrariness of conventions, their underlying purposes, and the attitudes of others to conventional practice, but it constitutes no obstacle to following them. They are very often deferred to by means of automated and conditioned reactions.³¹

Within this broad conception of mind, language, and meaning, Millikan presents her theory of conventions. As she points out, conventionality is composed of two related characteristics:

First, natural conventions consist of patterns that are ‘reproduced’ . . . . Second, the fact that these patterns proliferate is due partly to weight of precedent, rather than due, for example, to their intrinsically superior capacity to perform certain functions.³²

The conventional patterns reproduced over time form “lineages” (e.g., the lineage of the exclamation “Fire!” is the history of all its usages—applications of the tokens of the exclamation in concrete real-life situations). The reason that behaviors are reproduced over time is their proven effectiveness in bringing about the result(s) desired by cooperating partners (e.g., running away from the source of fire).

Millikan distinguishes between blind, half-blind, and open coordinations.³³ In blind coordinations, each partner has to act in ignorance of the other's actions, as in typical Lewisian examples of an interrupted telephone connection and road traffic. However, Millikan holds that the coordinations achieved by language are not blind coordinations, but half-blind or open coordinations, i.e., leader-follower coordinations:

Conventional leader–follower co-ordinations begin when a leader reproduces a certain portion of a traditional pattern, that portion being observable to a follower. The follower is familiar with the pattern, recognizes it, and reproduces the complementary part, resulting in a coordination of a sort that is partly responsible for the proliferation (due to precedent) of the pattern.³⁴

Millikan's “leader-follower” conventions are in no way to be construed as merely conventions that effectuate commands given to someone of inferior status (i.e., as being analogous to commands issued by, e.g., a colonel to a corporal). They include any convention that involves an interaction triggered by one person and followed by another, i.e., any call for action that elicits a positive response. A question asked and answered, a message conveyed by someone and then digested and used by someone else, and a request to open a window on a bus, followed by a kind fellow commuter doing so, all operate via a leader-follower convention. Commands and directive language obviously follow this pattern as well.³⁵

As shown below, the crucial issue for jurisprudence is that legal conventions are leader-follower conventions applied in open coordination situations where actions are preceded by communication. Nevertheless, legal philosophers have usually analyzed legal conventions by applying Lewis's ideas, premised as they are on blind coordination, and therefore treated all conventions as compliance-dependent. This approach is misguided and has had adverse ramifications for jurisprudence: it has led legal philosophers to argue that the RoR cannot be conventional as it is not compliance-dependent.³⁶ As shown below, Millikanian non-compliance-dependent conventionalism can be reconsidered as a theoretical framework for analyzing the conventional nature of the recognitional practice by post-Hartian jurisprudence.

III. LEGAL CONVENTIONS IN THE LIGHT OF MILLIKAN'S THEORY

A. Are Legal Conventions Rule- or Precedent-Based?

Is a convention necessarily a rule? Lewis's definition does not include the word “rule”: the genus used by Lewis is “regularity,” which is a synonym of “rule,” but only in its descriptive meaning (“as a rule, I commute by train”). “Regularity” does not cover the normative meaning of “rule,” which is one of the reasons for the criticism directed at Lewis.

Whether Lewis's conventions qua regularities are rule-based is therefore not clear. Postema describes the convention in Lewis's thought as “the ‘rule’ instantiated in those regular patterns of behavior and associated attitudes.”³⁷ Marmor explicitly defines conventions as rules (or norms) in his definition of conventions.³⁸ Coleman also claims that the convergent behavior of social actors applying conventions must be rule-governed.³⁹

For Millikan, conventions are essentially tokened precedents (“precedents in extension”) handed down from one pair of communicators to another.⁴⁰ Millikan compares the propagation of conventional forms to the propagation of genes, but in the realm of culture, and focuses on how users interpret the precedents that guide them during linguistic communication.⁴¹

Therefore, Millikan repudiates the notion that conventions, especially linguistic conventions, are rules.⁴² She claims that conventional regularities cannot be expressions of prescriptive rules, as they can be disobeyed without incurring sanctions, and cannot be ensconced in any objective medium.⁴³ If there are any conventional rules, they “describe conventional patterns; they do not prescribe them.”⁴⁴

Millikan is obviously aware of the possibility of generalizing from several prior instances of successful speaker-hearer coordinations, but contends that

a tendency for people to perceive certain patterns in the data might be widespread without being rigid, and without involving any “rules” subsisting either in the public domain or in individual people's minds.⁴⁵

She therefore proposes to treat conventions as precedent-based, regardless of how intuitive it might seem to regard them as rule-based. This approach does not require the formulation of general or abstract rules, as behaviors follow precedent by copying prior conduct into new situations.

François Recanati offers a clear presentation of how a precedent-based convention operates.⁴⁶ He argues that language can be applied in new circumstances without the intermediate step of forming a definition, by holistically comparing previous applications (“source situations: a collection of legitimate situations of application”)⁴⁷ with the current one (“target situation: the situation of current concern”).⁴⁸ This process, mutatis mutandis, can be used to demonstrate conventional linguistic behavior in which there is no need for the intermediate step of forming a prescriptive rule.

Recanati distinguishes between the “Fregean picture of concepts,” in which words are associated with abstract “conditions of application,” and an approach that associates linguistic types with their specific applications.⁴⁹ This distinction assumes that using language does not require the intermediate step of abstracting the common qualities of the previous uses of a particular word, as is done when formulating a definition.⁵⁰ The application of language is not an application of a predefined, narrow set of definitional criteria to the case at hand but rather a global comparison of the target situation with the source situations.⁵¹

Recanati's distinction between the source situations and the target situation is analogical to a distinction between previous situations in which a convention has been applied and a situation in which that convention is to be applied. When faced with a new situation, we do not generalize by abstracting a set of features from the previous instances of usage (forming a definition or a rule), but rather holistically compare previous situations with the new one, and if we find them relevantly similar, we behave as before. Our behavior is therefore governed by a precedent rather than by an abstract, general rule.

The assumption that socially recognized, stable, shared rules are not necessary for conventions to operate may suggest that Millikan's precedent-based approach is vulnerable to the objection that conventions not based on rules are unstable. That objection can be refuted by recourse to Millikan's stabilizing function concept, according to which the instability of the application criteria does not imply the instability of linguistic meaning.⁵² Millikan provides a convincing argument as to how linguistic practice leads to the emergence of a “stabilizing” or “standardizing” function for a particular linguistic device (e.g., a predicate), which forms “a center of gravity to which wayward speakers and hearers tend to return after departures.”⁵³

If Millikan's model of conventions is applied to law, then the rule of recognition, if it is to be perceived as conventional, should be seen as a series of acts of recognition, i.e., as reproduced responses to the lawmaker's communicative behavior, copied one from another and forming a historical lineage.⁵⁴ To explain this, we need to extrapolate Recanati's distinction between the source situations and the target situation to recognitional practice.

If recognitional practice is a lineage of copied behaviors, then it has to consist of particular acts where officials have recognized a valid rule and which have served as precedents for further similar acts. In linguistic terms, the act of recognition within this theoretical framework consists in applying the phrase “this is a valid rule” to a rule produced by the lawmaker. The phrase “this is a valid rule” has a “semantic potential,” which is the “collection of legitimate situations of application.”⁵⁵

The features shared by the source situations and the target situation, and consequently the conditions of application (e.g., of the predicate “P is a valid rule”), are not always the same,⁵⁶ as they depend on the target situation.⁵⁷ Recanati explains why this is so with recourse to Waismann's concept of open texture:⁵⁸ it is impossible to provide a complete definition of concepts, “a thought model which anticipates and settles once and for all every possible question of usage.”⁵⁹ The reason is that such a model would have to describe every potential element of a situation type in which the given concept is to be applied, and that is clearly impossible. Therefore, the definitions only comprise the pivotal qualities shared by the source situations (“the tip of the iceberg”), although the source situations are similar in many other respects (the “background”).⁶⁰ Whether the similarity between the source situations and the target situation will concern a feature mentioned in the definition or one belonging to the “background” depends on the target situation.⁶¹ A typical target situation (e.g., an uncontroversial statute enacted by a parliamentary majority), shares the components specified in the definition of a valid rule (i.e., the requirement of being enacted by a parliamentary majority)—an easy case. If, however, the target situation is atypical (e.g., a statute enacted by parliamentarians directly threatened by a military junta), its similarity to the source situations has to be assessed holistically, i.e., the similarity of the background features (e.g., duress or lack thereof) is also assessed.

Riggs v. Palmer provides an instructive case study.⁶² The dispute between the judges can be seen as a clash between the criteria-based and precedent-based approaches. For Judge Gray, the same criteria applied in similar prior cases should be applied. Relevantly, a last will that met all statutory requirements and had not been revoked should be deemed effective. Judge Earl's approach was more precedent-based in the Millikanian sense: whether the target situation (Palmer's will) was relevantly similar to the source situations (previous last wills) depended on more than a rigid set of criteria. This case differed from previous last will cases in that the beneficiary had murdered the testator. This element was not in the “tip of the iceberg” (narrow criteria), but it was a material fact that distinguished it from prior situations. This justified a conclusion that differed from those in the source situations.⁶³

Millikan's and Recanati's approach to linguistic practice sheds new light on the nature of the criteria used in the RoR. A corollary of the precedential nature of recognitional practice is that each subsequent act of recognition is performed on the basis of a holistic comparison of the target situation (a new legal rule) with the source situations (previously recognized legal rules), and not on the basis of a set of criteria derived by generalizing from previous acts. Seen from this perspective, the practice of recognition is not a practice of applying a set of criteria to the case at hand, but a precedent-based practice of recognizing as valid those rules that bear a relevant resemblance to previously recognized ones. This approach enables the criteria of recognition to be derived from the past practice of recognition; however, those criteria would only describe that practice and not normatively guide future practice. They therefore cannot be treated as prescriptive but only descriptive.⁶⁴

Hart seems to treat the criteria of recognition as prescriptive, not descriptive, and on this issue, Millikan stands opposed. However, Hart's conceptualization of language was by no means criterial, i.e., he did not believe that a sole binding set of criteria governing every conceivable case of language application could be formulated. Therefore, interpreting recognitional practice in the light of Millikan's conventionalism accords with Hartian anti-essentialism.⁶⁵ If there is a parallel between acts of concept application and acts of recognition, then Hart's conviction that using a general term in several instances is not based on any common qualities shared by those instances⁶⁶ can be paraphrased by claiming that the instances of recognizing a valid rule are not governed by a prescriptive rule based on a set of common qualities, but are governed by precedent.⁶⁷

Even if the Millikanian approach to RoR shares crucial features with customary law (e.g., it is created and changed not by what people say is to be done, but by what they do),⁶⁸ her precedent-based approach to RoR does not treat it as a typical customary rule.⁶⁹ This is not because the RoR is not customary, but because it is not a rule. It is a customary practice whose normativity is derived from it realizing its proper function (as shown below), not from being rule-based. The criteria applied within this practice cannot be encapsulated in a rule because they cannot be identified until the practice has been developed. Hart himself points to the primacy of the practice of recognition when he states that, in case of doubt as to the content of the RoR, “it could be established by reference to actual practice: to the way in which courts identify what is to count as law.”⁷⁰ Stanley Fish accuses Hart of putting the cart (the practice) before the horse (the rule(s) governing the practice), stating that “[i]f . . . one can know the rule only by extrapolating from practice, then practice rather than being generated or tested by the rule is the source of the rule.”⁷¹ This criticism is completely unjustified: the primacy of practice is not a fault but a feature of Hart's theory, when seen from the perspective of Millikan's theory of conventions.

B. Arbitrariness of Legal Conventions

Millikan concurs with Lewis's view that conventions are arbitrary, i.e., that every convention has a viable alternative.⁷² The conventional organization of traffic illustrates the point. Whether the right or left side of the road is chosen is arbitrary because a viable alternative is always available.

The difference between Millikan and Lewis concerns constraints on arbitrariness, i.e., the limits within which the community is indifferent between the available alternatives to a conventional behavior. As Lewis points out, a given coordination problem may have several solutions, each capable of achieving several coordination equilibria.⁷³ These equilibria can differ in quality, i.e., they can be Pareto superior or inferior.⁷⁴ The criterion for choosing one solution over another is the preference of the actors.⁷⁵

For Millikan, the constraints on arbitrariness do not depend on the actors’ mental states (their beliefs, intentions or—as in Lewis—preferences),⁷⁶ but on the ability of the convention to perform its function. In Millikan's terminology, the difference between Pareto superior and inferior equilibria can be expressed as the difference between two conventional behaviors that realize the proper function of the convention in a better or worse manner. She assesses the quality of conventions in terms of whether and to what extent they reflect the state of the world (in conventions whose proper function is to create proper beliefs about the world) and/or their effectiveness in changing behavior (in conventions whose proper function is to cause people to do things).

Arbitrariness is often linked with another feature of conventions, viz the underdetermination of values by the conventions that serve them, i.e., the fact that a value can be realized or served in a number of different ways.⁷⁷ At a certain stage of its development, a particular society may use a particular convention to achieve its purposes and values. If a better way of realizing those values presents itself, the society may switch to it. This switch would not primarily be the result of a change in individual preferences, but of the new convention being better suited to fulfilling its proper function, and therefore less arbitrary.⁷⁸

Perceiving the arbitrariness of conventions from the perspective of the underdetermination of values explains why there can be disagreement concerning the application of a particular convention (the actors may disagree on the best way of fulfilling its proper function). It also explains the change mechanism of conventional behavior (a better conventional solution, i.e., a better determination of an underlying value, incrementally gains traction in a particular society by being reproduced increasingly often, and gradually displacing the older one). Elucidating the arbitrariness of conventions with recourse to Millikan's notion of proper function therefore puts us in a better position to explain the possibility of disagreement and the change mechanism in conventions—phenomena not easily explained by Lewisian conventionalism. If, as according to Lewis, most members of a community have to follow a convention before it can qualify as such, then it would seem that any amendment necessarily involves its temporary absence, i.e., a shift from Convention A to Convention B necessarily involves a gradual departure from the former and an equally gradual arrival at the latter. There is therefore a period during which neither A nor B are universally followed.

If the constraint on the arbitrariness involved in choosing a convention is that the latter has to perform its proper function, the arbitrariness of the conventional practice of recognition also has its constraints. Some relate to the fact that the conventions involved in legal practice are leader-follower conventions, within which the leader has to be able to tell the follower what to follow. Promulgation fulfills that role in the legal system. Publishing a statute is an act of public communication in which the leader apprises the followers of mandatory acts and/or omissions. That promulgation is not optional and therefore not arbitrary is confirmed by Fuller, who includes promulgation in his list of eight features that a law has to possess to be law proper.⁷⁹

The Fullerian list details further limitations on the arbitrariness of legal conventions in general, and on recognitional practice in particular. Some of these necessary features concern the content of the law, not its form or promulgation. For instance, the Fullerian requirement that the law does not have contradictory rules or impose impossible obligations clearly concerns the content of legal rules. As those requirements are necessary, they cannot be arbitrary. Unless it meets those requirements, the law cannot perform its proper function of the leader-follower convention, as it cannot guide behavior. After all, nobody can respond to a conventional action that simultaneously requires doing A and not doing A (contradiction) or doing the impossible. If there is no possible response, then the convention will never be reproduced.

The above considerations lead to two important consequences of Millikanian conventionalism for our understanding of recognitional practice. First, recognitional practice is based on a convention that is not wholly artificial and arbitrary: it has to meet certain requirements in order to serve its proper function. Second, contrary to a belief that is widespread in jurisprudence, some of those requirements concern the content of the law, not its source or form.⁸⁰ Recognitional practice therefore has to recognize the content of legal rules. In other words, if recognition is a form of uptake that the law requires for its validity,⁸¹ then the followers (officials) are, in certain situations, obliged to not recognize the rules of some content as valid.⁸²

The arbitrariness of conventions need not be transparent to everyone involved in following them, as conventional behaviors may be copied habitually and nonreflectively through conditioning. The extent to which it is transparent can differentiate professionals from laypeople. The former are usually better informed about the available alternatives to a convention than the latter. This difference helps explain the respective roles of officials and laypeople involved in the conventional recognitional practice in law, as envisioned by Hart.

Moreover, the discussion on which features of legal conventions are arbitrary can be a source of theoretical dispute between conventional practice participants. And when a particular participant arrives at a better understanding of the arbitrary or nonarbitrary nature of a particular feature of law, he/she may attempt to shift recognitional practice in a new direction. If that shift is followed by others, there will be a change in conventional practice. Therefore, the new light that Millikanian conventionalism sheds on the arbitrariness of conventions facilitates understanding of how conventions—and by corollary, recognitional practices—evolve.⁸³

C. Compliance-Dependence of Legal Conventions

Millikan, contrary to Lewis,⁸⁴ argues that conventions need not be adhered to universally or almost universally to operate successfully, although blind coordinations (e.g., the conventions for redialing when a connection is broken or traffic conventions) may be an exception.⁸⁵ The conventions used in half-blind and open coordinations, however, do not require regular conformity.⁸⁶ These “leader-follower” conventions need only be adhered to in a critical mass of cases. Linguistic conventions are an example,⁸⁷ as are legal conventions, RoR included. As Millikan argues:

all that is required for a leader-follower convention to survive, to be repeated and passed on, is to succeed in coordinating the interests of speakers and hearers some critical proportion of the time, weighing the value of coordination successes against the disvalue of failures. . . . Regular successes are not needed to maintain a leader-follower coordination convention, any more than they are needed to maintain many biological traits. If the cat catches one mouse for every ten pounces, and the beggar receives one quarter for every ten people accosted, it is sensible for both to keep at it.⁸⁸

In other words, a precedent only has to fulfill its proper function in a sufficient number of cases to proliferate. Not deferring to it does not endanger the conventional practice so long as it makes the social actors better off overall.

It may be objected that the advantage of Millikan's model is dubious, as competing approaches can accommodate errors and defections from the conventional practice, provided that the practice itself is patterned enough for the convention to be discerned. It is important to stress that Millikan does not simply lower the threshold of compliance in conventions as compared with Lewis; she takes a completely different approach to compliance-dependence. Her theory does not require even general compliance for a convention to exist: it can apply to rare occurrences (as in the beggar example), so long as it delivers a proper function. Moreover, as it would be an oversimplification to say that biological evolution accommodates errors and defections (if anything, it is fueled by them), it would be an oversimplification to say that a robust theory of conventions only has to accommodate errors in the conventional practice. It has to explain how the defections help develop it. Compliance-dependence, which in Lewisian conventionalism is a metaphysical condition for the convention to exist,⁸⁹ is also a crucial theoretical barrier to explaining how conventions can change over time. Another advantage of Millikan's non-compliance-dependent approach is that her theory is free from that barrier.⁹⁰

In Lewisian conventionalism, the requirement for universal adherence is linked with the requirement for compliance-dependence or mutuality of conventions.⁹¹ Compliance-dependence, or the conviction that deferring to conventions is necessarily linked with the expectation that other social actors will do likewise, is perceived by many as a definitional feature of conventions:⁹² “the point of complying with the convention depends on the compliance of others—absent their compliance one does not have reason to comply.”⁹³

Millikan's approach to compliance-dependence is different: it is based on the local perspective of an acting individual whose knowledge of others’ intentions and preferences is limited, and for whom complex reasoning about other people's future behavior may be impossible or impracticable due to lack of time or ability. This difference is nicely encapsulated by a reference to Hume's metaphor, in which conventional arrangements in a society can be compared to a vaulted ceiling, “each stone having a place and depending on the others to accomplish their common task.”⁹⁴ The very act of perceiving the vaulted ceiling is only possible from an external standpoint from which the entire conventional arrangement is visible. This is the Lewisian perspective. Millikan's perspective is subjective and internal. It is the perspective of a particular stone in a vaulted ceiling or of an actor taking part in a project that he/she does not need to perceive holistically as a wider activity or plan.⁹⁵

Consequently, Millikan not only holds that conventions do not need to be adhered to universally, but that a competent social actor, despite having limited knowledge about other people's internal states and preferences, at least knows that they do not always defer to conventions. This renders any unconditional assumption about other people's behavior counterfactual. Any expectation of compliance is limited to a tentative expectation concerning the other party, i.e., the actor present here and now, and not the whole community. Contrary to Lewis and Grice, no endless doxastic cascade of expectations is in play.⁹⁶

For Millikan, the expectation of general compliance is not a factor in decision-making (i.e., a reason for answering “Should I follow that convention?” in the affirmative), but rather a more general result of a community's following useful conventions. For Lewis, conventions work because they are followed; for Millikan, they are followed because they work. The existence of a convention depends on the benefit that accrues to both leader and follower from deferring to it, not their mutual expectations that it will be followed unconditionally and without exception (i.e., even if the convention does not bring about the desired result). In other words, whether others do something is no longer a compelling argument for me to do likewise. The only valid reason for following the convention is that it performs a proper function and delivers a survival value. If those benefits induce a significant portion of society to follow conventions, widespread compliance is a consequence of those benefits, not their cause.

The fact that Millikanian conventionalism is not compliance-dependent saves it from most of the criticism raised against its Lewisian counterpart. The key point of that criticism has always been that legal conventions are not compliance-dependent like Lewisian conventions. Dickson convincingly argues that the RoR is not a conventional rule, as compliance-dependence is not among the reasons it is followed.⁹⁷ Shapiro, Toh, and Raz similarly base their criticism of conventionalism on its compliance-dependence.⁹⁸

Millikan's approach shows that a practice can be conventional even if it is not compliance-dependent. It is obviously not sufficient to prove, contra Dickson, that the RoR, as envisaged by Hart, is a conventional rule, as Dickson's arguments are compelling. Hart could not possibly have been familiar with Millikanian conventionalism (although her most important book on the philosophy of language was published in 1984, her works on conventions were published long after Hart's death). Therefore, any claim that Millikanian conventionalism could define the original Hartian RoR is at best speculation.

However, the fact that Hart could not have applied Millikan's theory to the practice of recognition does not mean that we cannot do so. Legal philosophers abandoned conventionalism primarily because of its compliance-dependence, and consequently, its inability to explain why legal rules are obeyed (because others do so can be discounted). Therefore, a conventionalist approach whose main tenet is not compliance-dependence offers a chance to revisit conventionalism, and paves the way for a post-Hartian approach to the recognitional practice, based on the initial Hartian intuition that sufficiently sophisticated conventionalism is able to explain how a social practice can produce law.

Millikan's position concerning adherence to conventions and their compliance-dependence helps answer several challenges to Lewisian conventionalism, one of which is critical, viz how it is possible for social conventions to be disputed by their users if they have to be universally adhered to.⁹⁹ This question has a crucial bearing on discussions concerning the recognitional practice.

If leader-follower conventions do not need to be adhered to universally, and recognitional practice is based on a convention of that type, then it does not need to be adhered to universally. This is a crucial difference between how recognitional practice is perceived by the Lewisian and Millikanian conventionalisms. Millikan does not regard the possibility of an individual refusing to follow a convention as a threat to its existence. On the contrary, it is an intrinsic, and indeed necessary, component of a living conventional practice, which—due to its evolutionary nature—is a process of trial and error.¹⁰⁰ Therefore, both leaders and followers can propose new solutions. Their sole assessment criterion is whether they promote or hinder the proper function of the convention.

A leader may decline to follow a convention by expressing a directive in a modified way, or by explaining the purpose of the directive to the follower before issuing it (not having done so previously). If these sorts of modifications make the follower more inclined to follow the directive, they will proliferate. On the other hand, the follower might refuse to follow a particular directive because following it would deprive society of an important benefit achieved by another, earlier directive. As a result, society may change the convention by allowing the followers to weigh the results of following one directive against another (a kind of state of necessity). These two examples show that society is not concerned that conventions remain intact, but that they are effective in bringing about their customary benefits. And conventions sometimes have to change to stay effective.

Millikanian conventionalism can help ease the “dialectical tension” between continuity and novelty in law and legal interpretation, signaled by Raz. Arguing that continuity naturally secures legal stability (“That things happen in a certain way makes it right, or good, that they should continue to happen in that way”),¹⁰¹ he illustrates this with the example of conventions, as the category “in which behaviour is justified if, and normally only if, a general practice exists.”¹⁰² This touches on the problem of the compliance-dependence of conventions. Raz also argues, however, that interpretative practice in law “lives in spaces where fidelity to an original and openness to novelty mix.”¹⁰³ Once more, it is extremely difficult to reconcile the compliance-dependence of Lewisian conventionalism with its openness to change, as the consensus required by compliance-dependence hinders innovation. In Millikanian conventionalism, by contrast, the dialectical tension finds its full explanation: the precedent-based conventional practice ensures stability, but the fact that it is not compliance-dependent allows for innovation and incremental change. In fine, Millikanian conventionalism is more compliant with Raz's expectations toward interpretative practice than its Lewisian counterpart.

Another challenge that Lewisian conventionalism seems unable to meet is that of Kutz, viz how can the fact that getting things right is often more important than getting them together be explained by Lewisian conventionalism?¹⁰⁴ Or as Postema formulated it: How can the phenomenon of “go it alone” in conventional practices be explained?¹⁰⁵ These challenges are inextricably linked with the crucial problem of how disagreements are possible in a conventional practice that by definition is based on universal adherence. As Lewis's commentators stress, even if such disagreements appear (as they must), they have to be made consilient.¹⁰⁶

According to Millikan, they do not have to. She removed the main obstacle to reconciling the existence of communicative conventions with individual defection therefrom when she demonstrated that they did not have to be universally adhered to. Once conventional practice is understood as being based on precedent, individual defections are like mutations,¹⁰⁷ i.e., they are either improvements or deteriorations in realizing the proper function and delivering survival value. If they are deteriorations, they are simply not reproduced, and therefore do not threaten the stability of the conventional practice. If they are improvements, they are reproduced because of their superior ability to bring about the proper function the conventional practice serves.

In Millikan's conventionalism, disagreement is not a “bug,” but an “undocumented feature,” which ensures that they work and will therefore endure. Precedent-based conventional practice is a process of trial and error, where the convention is constantly adjusted to suit the changing environment and new states of affairs. Millikanian conventionalism is based on the externalist concept of language and mind.¹⁰⁸ As such, it recognizes that any change in the external world or in our knowledge thereof may have a crucial impact on the content of our language and thoughts. As a corollary, that change can force a change in our conventions, if they are to remain effective tools in a changing world. The participants in the conventional practice, recognizing that a change has occurred in the world, are not only permitted, but in some sense obliged, to argue for a change to the conventions, if they care for the common good.

Therefore, to ensure that the convention delivers the usual benefits and concomitant survival value, it is sometimes necessary to “go it alone” (e.g., when faced with a novel situation). And it is always more important to get things right than to get them together. “To get things right” means to apply the convention in such a way as to ensure that its proper function is fulfilled. And this may require a change in the conventional behavior.

Because Millikanian conventions do not have to be universally adhered to, her theory is better suited than Lewisian conventionalism to answer the fundamental challenge of contestability.¹⁰⁹ Postema argues that it would be a serious mistake to think that, while legal practice in general is essentially discursive, i.e., consists in arguing which conduct is required by law, it is not so at its conventional foundations.¹¹⁰ A comprehensive account of legal conventions must acknowledge that the recognitional practice is not hermetically sealed from controversy, but is rather shaped by it.¹¹¹

Unfortunately, despite recognizing the fundamental adversarial nature of law, none of the accounts of legal conventions hitherto proposed have sought to explain it within a wider theory of conventions, and the commentators admit as much.¹¹² Jurisprudence is still waiting for a theory of conventions that can accommodate the essentially discursive aspect of law.¹¹³

Millikanian conventionalism successfully meets that challenge. First, it provides a theoretical space for differences of opinion within conventional practice, as universal agreement is not a necessary tenet.¹¹⁴ Second, the precedent-based nature of Millikanian conventional practice not only allows for, but necessarily requires, that the parties present arguments for or against the relevant similarity between the target situation and the source situations.

Millikan draws the inevitable conclusion from a fact that Lewis noticed but ultimately neglected, viz that no new instance of applying conventions is identical to the previous ones.¹¹⁵ If this is the case, then whether the features of the new situation in which a conventional behavior is to be copied are relevantly similar to the previous situations will often be subject to dispute. In cases of obvious similarity, the argumentative aspect of the practice may go unnoticed. It is, however, conspicuous in hard cases, when not only the similarity of the situations, but also the relevant features under which that similarity is to be assessed, are open to question. As a consequence, those who have adopted a given convention may differ both as to its requirements and its applicability to a particular case.

Millikanian conventionalism consequently proves that disagreement and conventions may not only peacefully coexist, but can symbiotically support each other. This claim is of the utmost importance to our understanding of the nature of the RoR. By accommodating disagreement within the theory of conventions, Millikan's conventionalism allows those aspects of rules on which lawyers disagree to be included in recognitional practice. A case in point are the rules of legal interpretation, which have been traditionally perceived as not belonging to the “criteria” of the RoR on account of their fundamental disputability.¹¹⁶ Millikan's redefinition of the compliance-dependence of conventions helps explain the “go it alone” phenomenon, i.e., the tendency to unilaterally withdraw from a convention.¹¹⁷ Millikanian conventionalism does not regard this tendency as necessarily reprehensible, not least because an individual may decline to follow a convention to strengthen, not weaken, the conventional practice by improving its ability to perform its proper function. In other words, defection does not have to be the result of an idiosyncratic need to apply personal preferences or criteria to the conventional practice.¹¹⁸ It may be caused by a completely sober decision that doing things differently will be better for society. Therefore, the convention users may argue for a change in the conventional practice, if they find that change is necessary to preserve the ability of the convention to perform its proper function.¹¹⁹

Therefore, in her response to Kutz's challenge to Lewisian conventionalism (why getting things right is more important than getting them together),¹²⁰ Millikan might well have argued that it is definitely more important to get things right than to get them together. After all, the real value of conventional practice is not about doing what others do, but doing the right things, i.e., the things that put the participants in a better position vis-à-vis the world they are navigating.

By answering the challenge of contestability, Millikan's conventionalism lays the groundwork for a definitive rebuttal of the Dworkinian challenge to legal positivism, in particular his claim that immanent contestability in law is inconsistent with it.¹²¹ That claim is true with regard to theories based on the Lewisian account, as they inherit all the problems which that kind of conventionalism encounters, in particular its compliance-dependence, and therefore became hereditarily vulnerable to the challenge of contestability. Millikan's conventionalism is impervious to this challenge and therefore better suited to strengthen the conventionalist justification of legal positivism.

D. Necessity of Ensuring Coordination in Legal Conventions

Lewis perceived conventions as solutions to the coordination problems that societies face. As with the other elements of his theory, this claim has drawn criticism.¹²² Conventional behaviors may have other functions, e.g., producing a mental representation in another individual or expressing an emotion.¹²³ Millikan's approach allows for a greater variety of benefits accruing from conventions, and thus obviates the problem signaled by Marmor, viz that constitutive conventions cannot be easily accommodated within Lewis's concept of conventions, as they do not solve coordination problems.¹²⁴

Millikan concurs that some conventions have a proper function other than coordination, e.g., the conventions of fashion can have a proper function of signaling membership of a particular group, manifesting status, showing respect, etc.¹²⁵ Therefore, Marmor is correct in his assertion that not all conventions ensure coordination. The direct proper function of some conventions is to elicit a mental representation or a combination of mental representations.¹²⁶ Whether these mental representations are used for coordination or for some other purpose is another issue.¹²⁷

Another essential difference between Lewisian and Millikanian conventionalisms is their diametrically opposed assumptions on whether coordination preceded communication or vice versa. Lewis assumed that coordination must have come first, because establishing a language would have required nonlinguistic (hence noncommunicative) conventions. Millikan perceives language and communication in the much broader sense of deriving information from the environment. She doesn't differentiate between deriving information from conventional signs and natural signs (e.g., between hearing someone say “it will rain” and seeing dark clouds gathering overhead).¹²⁸ With this broader understanding of communication processes, any coordination is dependent on a prior derivation of information, hence on communication sensu largo.

Millikan's leader-follower conventions are similar to a subset of conventions that Lewis called “two-sided signalling conventions,” within which “the coordination needed is coordination between communicator and audience.”¹²⁹ In some situations, these conventions are one-sided, e.g., when the role of the leader is not perceived, but the signs left by him or her are (Lewis's example is traffic lights). In a one-sided signaling convention “the coordination needed was either between communicators or between members of the audience.”¹³⁰

The question of the identities of the leader and follower, i.e., between whom such communication is to take place, naturally arises.¹³¹ In linguistic conventions, the leader is the speaker and the follower the hearer. The former performs the first part of the convention by describing the world or issuing a directive, and the latter responds by forming a belief based on that description or by complying with the directive. What are the identities of the leader and follower in legal conventions, especially in the RoR?¹³² Whose actions are coordinated by it?

The traditional jurisprudential approach holds that recognitional practice is a one-sided convention in Lewis's terminology, i.e., a convention that ensures coordination between officials as audience members.¹³³ The interaction within recognitional practice is therefore horizontal.¹³⁴ In this model, the officials are both leaders and followers, e.g., the judges who applied legal conventions in the past are leaders and the judges who apply them now are followers. This, however, is counterintuitive to Millikan. Actually, recognitional practice is a two-sided convention: one that ensures coordination between a speaker or leader (lawmaker) and an audience (officials and laypeople).¹³⁵

The leader-follower relationship in law is vertical, and connects the lawmaker (top) with officials and laypeople (bottom). The coordination between laypeople or between officials is a by-product of the coordination between the lawmaker and those two groups. They respond similarly to the directives issued by the lawmaker, and the coordination between them is a natural consequence of obeying the lawmaker's directives.

It is easy to show that the traditional focus on the coordination between officials is misguided.¹³⁶ Contemporary societies naturally have many officials involved in the recognitional practice. It is, however, not hard to imagine a society with a lawmaker, who issues directives to several laypeople, and with a single official, viz a sole judge responsible for recognizing and adjudicating on the directives issued by the lawmaker. Would a recognitional practice operate in such a society?

I believe the answer has to be “yes.” Recall that the main reason for having a recognitional practice is to mitigate the risk of uncertainty as to which rules are valid legal rules as opposed to the rules of morality or etiquette. There is no doubt that a society governed by many rules of different origin will encounter situations where the status of these rules is doubtful, regardless of whether that society has one or many officials. In this example, the sole judge does not need to coordinate his/her behavior with other officials, but his/her practice of recognition is nevertheless conventional. The judge frequently reproduces his/ her behavior of recognition in response to the rules issued by the lawmaker, and can do this by the weight of precedent. If the convention coordinates anything, it is the behavior of the judge vis-à-vis the lawmaker, not other officials (as in this thought experiment, there are none). The judge is a follower in a convention that has a “leader-follower” structure.

The reason why jurisprudence has focused on coordination between officials is simple. Legal philosophers automatically adopt that part of the Lewisian model intended to explain the conventions applied in blind coordinations, not open coordinations. Legal officials, however, are not separated from each other or from the lawmaker, with no possibility to communicate, as are car drivers and the parties to a broken telephone connection. Therefore, the coordination ensured by recognitional practice has to be understood as depending on the standardized behavior of followers in response to the actions of a leader or—in law—on the standardized response of laypeople and officials to the rules of the lawmaker.

Thanks to its two-sided structure, the leader-follower account of conventional practice, when applied to legal conventions, recognizes that this practice may include many and varied participants, both skilled (officials) and unskilled (laypeople). As such, it answers Postema's challenge of elitism, i.e., the requirement that a theory of legal conventions accounts for “outside-the-core participants,” those who are not members of the official “law-elite.”¹³⁷ The account proposed by Millikan definitely widens the scope of conventional practice by including the general population in “the recognitional community.” This account is close to Hart's original approach, and contradicts subsequent developments of his concept of the RoR on the part of analytical legal philosophers, who banished laypeople from the realm of conventional legal practice.¹³⁸

E. Normativity of Conventions in Law

One of the most often criticized features of Lewisian conventionalism is its inability to explain the normativity of conventions.¹³⁹ According to Lewis, the main reason that social actors follow conventions is that others do. An obvious problem with this explanation is that it is circular and that it derives an OUGHT from an IS: the reason for following the conventional rule is the fact that others follow it.

In Millikan, the normativity of conventions is connected with the notions of proper function and survival value, i.e., their ability to enhance the fitness of an individual or group. “Fitness-enhancing” and “survival” do not sound like typical normative terms. As Millikan convincingly argues, however, “normative terms are not always evaluative, but can indicate any kind of measure from which actual departures are possible.”¹⁴⁰ Eaton calls this sense of normativity “naturalistic and non-moralizing” and links it with a performance of some activity that “is proper to a thing.”¹⁴¹

Conventional practices “are concerned with achieving, realizing, respecting, or promoting certain values, principles or aims (individual or collective).”¹⁴² How do these values, principles, and aims relate to fitness or survival? The survival value generated by a convention is not to be understood as a direct contribution to an individual's biological fitness. A linguistic convention most often impacts survival indirectly, by enabling individuals to gather information and coordinate their interactions. For instance, the conventions that require that promises be kept, or clean shoes worn, directly generate a value that appears far removed from survival. After all, nobody dies as a result of not cleaning their shoes. These conventions, however, help build mutual trust, thereby shortening the odds of achieving other values more directly linked with survival. Causal links between a particular convention and its ultimate survival value often remain nontransparent to the social actors involved. Children are simply taught to be polite to each other and to defer to age. The rest is taken care of by an intricate network of interrelated proper functions that have emerged from a reproduced structure of conventions developed in society.

Millikan's normativity of conventions can be understood in the classical sense, i.e., in terms of conventions’ ability to provide reasons for action. The will to realize the proper function and thereby bring about a survival value is definitely such a reason, even if the actions of the convention users are the result of habit and conditioning. Millikan's approach to conventions is therefore not open to the charge of deriving “ought” from “is.” In contrast to Lewis, she does not claim that people ought to defer to conventions because others do so, but because the conventions have survival value. In fulfilling proper functions, they make people better off by helping them navigate the world and achieve their goals.

The normativity in Millikan's conventionalism has a biological dimension. She argues that the mental representation the leader evokes in the follower by using a linguistic convention is necessarily a pushmi-pullyu representation (PPR): “two-faced representation telling both what the case is and also what might be done about or with it.”¹⁴³ This means that the content of the mental representation evoked by the linguistic convention is both descriptive and normative: it presents the follower with “affordances,” understood as opportunities for actions to achieve valuable ends.¹⁴⁴

Millikan's biological explanation of normativity may seem odd to those accustomed to discussing the normativity of conventions in terms of values. Millikan's approach, however, is by no means idiosyncratic. Meredith Williams explains how we are biologically programmed to follow conventions “blindly” or conditioned to do so through acculturation. Following conventions in normal circumstances becomes second nature, tantamount to acting “with right but without justification.”¹⁴⁵ Millikan's account of normativity in which the social ultimately serves our basic biological purposes of survival and reproduction is similarly naturalistic, but not reductive.

The reason to continue applying a convention, viz the prospect of benefiting from the survival value that it has hitherto generated, is biological in nature:¹⁴⁶ “the normative element that is involved when one means to follow a rule is biological purposiveness.”¹⁴⁷ Copying behavior (here tantamount to following a convention, both by leader and followers) is intended to realize biological aims; it fulfills functions in accordance with evolutionary design.¹⁴⁸ These biological purposes and functions that constitute our reasons to act may be both innate and learned.¹⁴⁹

Achieving biological goals through behavior is very often based on following what Millikan calls a “proximal rule,” which ultimately leads to following a “distal rule.”¹⁵⁰ In the course of their evolutionary history, organisms acquire the ability to follow simple and easy-to-follow (proximal) rules that—if conditions allow—ensure that those organisms defer to more complex and more difficult-to-follow rules. Millikan's example: if a rat is sick within hours of eating something, it will never eat anything with a similar taste again.¹⁵¹ The proximal rule “If ingesting a substance is followed by illness, do not ingest anything with that taste again” is easy to follow.¹⁵² This rule is a biological means to following the more distal rule: “Do not eat poisonous substances.”¹⁵³ Obviously, following the proximal rule does not necessarily lead to following the distal rule—eating cheese and getting sick for an unrelated reason may cause a rat to follow the proximal rule without following the distal one, in which case it would fail to achieve the proper function of the whole mechanism (cheese is not poisonous). The idea behind the “proximal-distal” distinction is, however, that a normative biological mechanism can lead to a simple rule being followed automatically, which in turn leads to another, more complex rule being followed unconsciously.

The “proximal-distal” distinction can be useful in analyzing the normativity of conventions. It can safely be assumed that we follow many conventions out of habit or conditioning, and that the reasons for this are not fully transparent to us. Nevertheless, following such conventions often enough leads—if circumstances allow—to realizing more distal functions and purposes. Take the convention of obeying someone in authority. Our reaction to a command may be based on a PPR of the state of affairs we are to bring about as a result of the command or on conditioning that has made us automatically obey the commands of someone in authority. It is safe to assume that law, being based on a set of leader-follower conventions, began with individual leaders issuing ad hoc commands to individual followers. Those commands, like all linguistic tools, exploited the innate human mechanism(s) to respond to directive language. Thus, the normativity of simple directives is a result of our natural, and often involuntary, predisposition to create PPRs in response. In terms of the proximal-distal rule distinction, the reaction to simple directives is a result of following the proximal rule: “Do as you're told.” This easy-to-follow rule is a tool for following the distal rule “Coordinate your behavior with others.” Realizing the distal rule and obtaining its survival value are not guaranteed by following the proximal rule (the command may be unreasonable or impossible to obey). Generally, however, the biological normative mechanism that makes humans respond to commands contributes to achieving the more distal purpose of coordinating their actions.

Those legal positivists who accepted the Lewisian account of conventions encountered a serious problem with normativity, viz that the Lewisian conventions are factual regularities. As such, they cannot provide social actors with genuine reasons to act.¹⁵⁴ Millikan, however, holds that a conventional practice is adopted (consciously or unconsciously) to reproduce behaviors that have generated a survival value in the past. The challenge of normativity is met when Millikan's theory is transplanted to law: legal conventions should not be followed because others do so, but because they have survival value, thereby enhancing the fitness of individuals and the groups of which they form part.

The precedent-based nature of a conventional recognitional practice implies that following it is tantamount to deferring to previous acts of recognition. People copy the behavior of their predecessors when the target situation is relevantly similar to the source situations, and refuse to do so otherwise. Previous behaviors are not copied because others do so, but because the conventional practice generates a survival value to the society and the individuals that comprise it. In this sense, those who continue the practice have a duty to treat similar situations similarly and different ones differently. The conventional recognitional practice is normative because the onus is on those not following it to justify their refusal (the “comply or explain” principle).

The “proximal-distal rule” distinction can be also applied to recognitional practice. Here the proximal rule would be “Recognize rules that are X as valid.” Following this proximal rule would lead to following the distal rule “Ensure that the legal system X is coherent and effective.” Obviously, following the proximal rule does not guarantee that the distal rule will be followed. In the critical mass of cases, however, it has, and so we continue to operate according to that proximal-distal formula.

The “proximal-distal” distinction reveals another difference between officials and laypeople, and their respective roles within recognitional practice. Owing to their expertise and experience, officials are usually in a better position to comply with the distal rule, and they assess the proximal rule in terms of its ability to improve our chances of complying with the distal rule. Officials may be better “equipped,” both biologically and through their training, to see the bigger picture, e.g., exhibit specific cognitive skills, such as long-range planning and constructing complex mental representations.

Millikan's theory can accommodate yet another criticism against which Lewisian conventionalism is defenseless, viz that of Green, who argues that coordination conventions lack the sort of normative force that law claims.¹⁵⁵ His argument is that conventionalism can explain normativity only when there is no radical conflict among social actors over the mutual benefit the conventional practice delivers.¹⁵⁶ Green clearly assumes, following Lewis, that the preferences concerning the equilibria are sufficiently transparent to constitute reasons to follow or depart from the conventions. Therefore, Lewisian conventionalism, which is based on social actors’ preferences and agreement, cannot explain why the conventional practice, and not agreement and the fulfillment of preferences, should guide action. If this is so, the conventional practice is not the exclusionary reason for action. As was shown, however, in Millikan, the actors’ preferences and their agreement play no role in constituting the conventional practice. Moreover, according to her, the purpose of the conventional practice and the benefits it brings to society can be nontransparent to the actors, as the proximal-distal distinction clearly shows. It is extremely difficult to have preferences for, or radical conflicts over, things that are not transparent. Nor can an action be intended to achieve a purpose that is not transparent or even known to the actor. The only remaining possibility is that, within the proximal-distal formula, the proximal rule (e.g., “Do as you're told!”), is the only reason for action for the participants of that practice, and therefore the condition Green sets for the conventional practice being normative is fulfilled.

IV. CONCLUSION

The pundits of legal conventionalism believe that its prospects are promising, but not yet assured.¹⁵⁷ Millikan's theory of conventions has the potential to change that. However, the question remains: Why another conventionalism? Hart's intuition that sufficiently sophisticated conventionalism can parry the Dworkinian attack is eminently plausible. This paper has attempted to show that Millikan's conventionalism is better suited to achieving this end than its Lewisian counterpart, especially in view of its ability to counter normativity and contestability challenges. Millikan's conventionalism deserves to be better known in jurisprudential circles, especially as its tenets do not assume compliance-dependence, which has hitherto been the main target of anti-conventionalist criticism.

Millikan's notion of proper function can help defend legal positivism against more recent assaults at its foundations, specifically that of Greenberg. His key argument is that value-based reasoning is required to transform social facts into legal content, and that this reasoning is based on values external to the social practice that constitutes the basis of law.¹⁵⁸

In discussing possible objections to his theory, Greenberg identifies one according to which he relies on “too thin a conception of law practices,” and that “properly understood, law practices can themselves determine the content of the law.”¹⁵⁹ As he says, that objection “claims that the additional substantive factors are part of law practices themselves.”¹⁶⁰

Millikan's conventionalism allows for such an objection. Greenberg frequently presents a “practice” as a series of social facts.¹⁶¹ Millikan's concept of practice as a lineage treats “practice” as a series of social facts linked by the proper function they serve. In this sense, Greenberg's concept of practice is indeed too thin. At the same time, the proper function as a value-based factor is internal to, not independent of, the practice. That proper function is internal to the practice can be exemplified by using an artifact. Is its purpose internal or external to the practice of using it, e.g., is cutting bread internal or external to the practice of using a knife to cut bread? Is the knife's cutting ability external to the practice of cutting with it? I would say that it is an internal and indispensable part of the practice, not an independent factor, and that the law practice is the same. Its proper function is not independent from this practice, but is its defining factor. As Greenberg argues, “An internal-value theorist must explain how legal content determined exclusively by law practices and internal value facts can provide genuine, as opposed to merely internal, reasons for action.”¹⁶² As shown in Section III.E, the proper function provides legitimate reasons for those who follow the practice to act. Finally, proper function is a value-based but not necessarily moral factor—the conventional practice of using a guillotine for capital punishment realizes its proper function (whatever it is), but this has nothing to do with the morality or otherwise of the conventional practice of capital punishment.

According to Millikan, the value factor that transforms social facts into legal content could be, however, derived from within the practice: the proper function is such a factor. A series of independent behaviors is transformed into a purposeful practice because they serve the same proper function, and this allows the relevant and irrelevant behaviors within the practice to be distinguished. As such, the proper function is constitutive to the practice: it makes the practice what it is. It is also internal to the practice: the essence of the practice cannot be found outside it, and the proper function does not exist unless the practice does. Social facts are transformed into a law practice because a proper function is attributed to what has been said and done, and these words and behaviors are then reproduced because they serve this proper function. In this sense, it is the proper function that justifies treating some social facts as relevant to the content of the law and others as irrelevant.¹⁶³

The usefulness of Millikan's theory is obvious once we assume (as Greenberg does) that the model of transforming social facts into legal content is a theory of interpretation. In this sense, disagreement among judges in such cases as Tennessee Valley Authority v. Hill can be seen as a clash of two models of interpretation. For Greenberg, the value factor that enables the best interpretation model to be determined, (the “X” factor) is morality,¹⁶⁴ as it constitutes the normative aspect of the practice. That normative aspect does not need to be moral, however: it may be derived from its proper function, which does not need to have anything to do with moral norms, but can constitute the X factor. One can imagine a recognition practice that realizes a proper function because it ensures a proper transmission of preferences from the people to the law, and therefore does not cause social unrest, without assessing the morality of those preferences. Realizing the proper function may have a clear coordination purpose without any moral aspect.¹⁶⁵

Its ability to refute anti-positivistic criticism, its methodological rigor, and its ability to avoid the shortcomings of Lewis's conventionalism make Millikan's conventionalism an attractive theoretical framework. My hope is that this paper will help Millikan's work receive the recognition it deserves among legal philosophers.