Much of the world that we inhabit relies for its order, stability and familiarity upon rules and understandings that we hold in common but mostly take for granted. We might loosely speak of such things as matters of ‘convention’. But the single word ‘convention’ conceals the diversity and complexity of the phenomena in question. In some contexts the notion of ‘convention’ is associated with agreement, and ‘agreement’ is itself ambiguous as between a convergence of opinion (e.g. we agree that the Moon is not made of cheese) and the exchange of promises (we agree to meet next Saturday). Thus we speak of ‘conventional beliefs’ but also of ‘the Hague Convention’. In other contexts ‘convention’ is not connected with agreement at all, but with shared standards or patterns of conduct that derive a part of their practical force from the very fact that they are shared and regularly complied with.

It is conventions in this latter sense that form the subject matter of Marmor’s carefully argued and enlightening book. The opening three chapters develop a view of conventions which is then applied to language, morality and law in the remaining four chapters. The book is written in a pleasingly concise style, but this means that its slender volume belies its complexity. Since it would be neither possible nor helpful to summarise the argument of the entire book, I will concentrate on the general analysis of the first three chapters, an analysis which underpins the book as a whole.

The starting point for Marmor’s analysis is provided by an understanding of practical reason. Practical reasons are ‘closely tied with the idea of value’: one has a reason for action insofar as the action is ‘good, valuable, or serves some worthy end’ (p. 5). Conventions are social rules, and the first step in understanding conventions must therefore be to understand the role of rules within practical reasoning. Rules are neither generally accepted reasons nor mere regularities of behaviour: rather, the basic function of rules is to replace first-order reasons for action, thereby avoiding the need for separate deliberation on each occasion. Conventions are rules that satisfy the following conditions (I simplify slightly): (i) the rule must be generally followed, in certain circumstances, within a certain group of people; (ii) there must be a reason for members of the group to follow the rule in those circumstances; (iii) there must be at least one other rule that could have been followed in those circumstances such that, had such an alternative rule been followed, the same reason that supports the current rule would have been a sufficient reason for members of the group to follow the alternative rule, at least partly because the rule is generally followed.

These conditions (especially the last condition) explain our sense that conventions are in some sense ‘arbitrary’, and Marmor takes this arbitrary character to be the general clue we need to decipher the unity underlying the seemingly endless diversity of conventions. Thus he tells us that ‘a practical domain that is conventionally determined is one that could have been different from the way it is, without any significant loss in its point, purpose, or value’ (p. xi). This insight, however, presents us with a challenge, for we need to explain how something that could easily have been otherwise, without any significant loss of value, can nevertheless feature in our practical reasoning as a guide.

Marmor emphasises that, if we are to understand the role of conventions in practical reasoning, we need to avoid confusing ‘arbitrariness’ with ‘indifference’. A rule may be conventional provided that the reasons supporting it might equally have supported some other rule (i.e. the rule is arbitrary), even if people have preferences between the alternative rules (i.e. they are not indifferent between the rules): what is essential is that the reason to prefer one of the potential rules must not be stronger than the reason to conform to whichever rule is generally observed (pp. 8–10).

We might at first be tempted to think that the notion of a convention can be understood by reference to the idea that general compliance with a conventional rule is necessary to the practical force of the rule. This is, however, in itself inadequate as a criterion of conventionality. As Marmor points out, general or widespread compliance may sometimes be a condition of a rule’s reason-giving force without the rule being conventional: I may have no reason to observe a prohibition on smoking in a certain place if no one else follows the rule and the place is already full of smoke. Yet such a prohibition is not a convention. The reason for following the rule in the first place is to avoid causing harm to others, and the desirability of avoiding such harm is quite independent of the compliance of others. Their non-compliance renders the objective impracticable in the circumstances, and their compliance is to that extent a condition of the practical force of the rule. But the point of the rule is not the establishment of a pattern of convergent conduct.

An influential analysis of conventions was offered by David Lewis in his book Convention: a Philosophical Study (Reference Lewis1969), but Marmor considers the analysis inadequate to cover the full range of relevant phenomena. For, on Lewis’s account, conventions are solutions to recurrent co-ordination problems, taking the notion of a co-ordination problem in a strict sense: a co-ordination problem involves actors facing a number of available courses of action in circumstances where each actor’s strongest preference is to act in concert with the other actors, rather than to pursue one or other course of action independently. Marmor points out, however, that Lewis’s analysis fails to fit some activities that we think of as governed by convention. Chess, for example, is not plausibly thought of as a solution to a co-ordination problem. Lewis’s analysis seems to assume that one first has a recurrent co-ordination problem and then a convention emerges to solve it. But there is not and never was a co-ordination problem to which chess is the solution. One could try to construe the notion of ‘co-ordination problem’ in an increasingly contrived and abstract way, but this is unlikely to be helpful or enlightening. It is better, Marmor judges, to abandon the idea that all conventions are solutions to recurrent co-ordination problems. The more general idea of ‘arbitrariness’ (relied on by Marmor) is therefore a better guide than the narrower idea of a solution to a co-ordination problem (relied on by Lewis).

Having in this way detached the notion of ‘arbitrariness’ from the context of co-ordination, the author is enabled to develop a notion of ‘constitutive conventions’. Adopting but modifying John Searle’s familiar distinction between ‘regulative’ and ‘constitutive’ rules (Reference Searle1969, pp. 33–42), Marmor tells us that constitutive rules are rules that constitute a social practice. Such rules can be either institutionally enacted rules or social conventions. Practices such as games, forms of art, ceremony and etiquette are constituted by social conventions that are quite distinct from co-ordination conventions. This analysis enables us to understand how chess and a host of other activities can be conventionally structured practices without being solutions to pre-existing co-ordination problems. But, perhaps more importantly, it also enables us to understand how conventions can give rise to the particular values that apply within a certain practice. Some of the values and standards of appraisal applicable to a theatrical performance, for example, are, according to Marmor, unique to the theatre, and created by the conventions that constitute the practice of theatre.

A practice, of course, is not exhausted by the rules that govern it. The practice could not exist without the rules, which in that sense constitute the practice, but ‘there is typically much more to a social practice than following its constitutive rules’ (p. 40). A conventional practice is not wholly detached from the background of human needs and aspirations. Indeed, the practice can only be understood in relation to more basic social needs and aspects of human nature that give rise to such practices, even though these features radically under-determine the nature of the practice (it is this radical under-determination that renders the practice conventional). Marmor acknowledges that ‘it is not easy to draw the distinction between values that are constituted by the relevant conventions and values that we associate with an activity just by grasping the nature of the activity and the kind of thing it is’. But we can say that ‘a certain evaluation is (partly) constituted by conventions if it is the case that we could not have grasped or come to appreciate the relevant value without understanding the conventions that constitute the relevant practice’ (p. 38).

Furthermore, ‘once a conventional practice is in play, the practice may constitute further values that can only be instantiated or appreciated by engaging in that practice’ or, to some extent, by non-participatory reflection upon the practice (p. 41). Indeed, social practices can embody an entire ‘grammar’ of evaluative concerns that can affect the point and ultimately the content of the constitutive conventions themselves. Thus, ‘constitutive conventions tend to be in a constant process of interpretation and reinterpretation that is partly affected by external values, but partly by those same values that are constituted by the conventional practice itself’ (p. 48). Social practices can exist and flourish even when the participants have competing conceptions of the values that the practice instantiates (p. 54).

The issues Marmor addresses at this point connect fruitfully with the work of such philosophers as Alasdair MacIntyre (who has written of goods that are internal to particular practices) (Reference Macintyre1981, ch. 14) and Joseph Raz (who has explored the possible dependence of values upon practices) (Reference Raz2003), as well as with broader efforts to ‘de-sublimate’ some or all of our values (see Simmonds,Reference Simmonds2007, chs 1 and 5). But the author wisely avoids a full-scale engagement with such large questions, preferring to concentrate on the development of a clear analytical framework that will assist in the formulation and elucidation of the relevant problems. Faced with a choice between epistemic and ontological interpretations of the dependence of some values upon convention, he simply notes that it is not necessary for him to resolve the question raised by that choice (p. 40). Nor does he see it as his task to address questions in the metaphysics of social reality: thus he sets on one side Searle’s (Reference Searle1995) claim that some notion of collective intentionality is necessary to enable us to understand the relationship between social facts and ‘brute facts’ (p. 54). Marmor wisely avoids the temptation to go further than he needs to, and in this way he increases the general value of his analysis.

In many contexts, the conventional status of a domain of human practice is philosophically contentious. Yet, as Marmor points out, it is somewhat odd that this should be so: we might expect it to be quite clear whether the relevant norms are conventional or not (p. 58). The explanation, he suggests, lies in the existence of ‘deep conventions’. These are conventions that emerge in response to basic needs and that enable surface conventions to emerge: the deep conventions are, in general, observed by following their corresponding surface conventions. Thus, to play a competitive game, participants must share certain deep conventional norms concerning the nature of games. It is the depth and complexity of human conventions that often obscures their conventional nature.

The author acknowledges the problems posed by the idea of ‘deep conventions’. For example, we may want to say that Islamic ornamental art and Christian representational art rest on different conventions. But, to apply the concept of ‘convention’ in this context, we must be able to preserve the thought that the norms are ‘arbitrary’ (though not, remember, ‘indifferent’). To preserve that thought, we must be able to view the differing norms as simply diverse ways of achieving basically the same thing. Thus we are told that the notion of convention remains useful in this context:

Yet it is very hard to say whether Christian art and Islamic art are different ways of doing the same thing, or achieving the same purpose: the question seems to lack a determinate answer, and may even lack sense. Even when a verbal formula is available with which we can ascribe a shared aim to Christian and Islamic art (e.g. Marmor suggests, ‘glorifying God’), we may well feel that the formula conceals discrepant worldviews rather than revealing genuinely common purposes.

At this point we may well begin to have doubts about the soundness of Marmor’s proposed extension of the notion of convention. The thought that art is structured by conventions is, of course, a very familiar one. Yet can that thought be fitted without loss to the template proposed by Marmor’s analysis? Does the analysis not threaten to have a flattening and impoverishing effect upon our comprehension of art? Christian artists may use blue to represent the Virgin, and we are at first inclined to think of this as a coded way of referring to the Virgin, or identifying a particular figure as the Virgin. But might another interpretation not offer greater insight? Perhaps the use of that particular colour is intended to alert us to connections with other manifestations of blue, thereby tracing resonances of meaning that exist quite independently of the conventions of art (consider the blue of the sky, the blue of the far horizon and the blue of the infant’s eye)? Of course, we must not confuse ‘arbitrariness’ with ‘indifference’: the use of blue may be a convention in Marmor’s sense even if there are reasons to prefer blue to some other colour, provided that the reasons favouring blue rather than some other colour are less strong than the reasons favouring conformity to general practice. But our doubts may be too fundamental to be resolved by this simple observation. For is the artist employing a system of conventions that facilitate reference, or is the artist exploring the infinitely complex resonances and relations of significance that give structure to a divinely ordered world? To insist that choice of colour is a coded way of identifying (as the Virgin) a figure in a painting is to beg the question of what the point of the painting is, of how it serves to ‘glorify God’, and therefore of why the artist wishes to refer to the Virgin at all. Might the glorification of God not consist precisely in a depiction of the subtle harmonies across space and time that give meaning to all that occurs within God’s creation? If we treat the choice of colour as a coded form of reference, do we not obscure precisely this aspect of the painting? Marmor emphasises that people may be observing conventions even when they do not understand that that is what they are doing (pp. 5, 6). But the question here is whether the artist’s choice of colour is simply a way of identifying one figure as the Virgin, or a way of suggesting the place of virginity (and of the immaculate conception) within the divine ordering of the world. If the latter is the case, the artist is inviting us to connect the thought of virginity with the blue of innocence and of infinity, and for that purpose no other colour will serve: ‘arbitrariness’ is absent and the practice is not a convention (in Marmor’s sense).

We may decide to reject extensions of David Lewis’s analysis of convention on the grounds that such extensions have to inflate the notion of ‘co-ordination’ beyond acceptable or enlightening limits: as Marmor points out, the game of chess is not a solution to a co-ordination problem. But is something similar not happening here to Marmor’s own analysis? Does it really make sense to think of the conventions of art as enabling the artist to achieve an independently describable object (such as the glorification of God)? Might this not obscure the extent to which the work of art expresses something that simply cannot be expressed in any other way, or understood apart from the particular expression that it receives? Analogous questions apply to the relationship between values and social practices: might a practice not provide us with the only way in which a certain value can be understood, rather than merely one of a diversity of ways in which the value may be realised? And might that not be something quite different from the value being constituted by convention?

It would, however, be an error to construe this as a reason for rejecting Marmor’s analysis, as opposed to his particular illustrations of that analysis. After all, in contesting the applicability of Marmor’s notion of convention to some aspects of art, we are applying his analysis; and the application of that analysis serves here to alert us to questions that we might not otherwise have been able to formulate. In the absence of the analysis, the possibility of such questions would have been obscured by the many ambiguities underlying the word ‘convention’. For, on either interpretation of the artist’s use of blue, the appropriateness of using blue for the Virgin might be described as ‘conventional’: either a conventional rule prescribing blue as a way of referring to the Virgin, or a conventional belief (see my opening paragraph) to the effect that the particular range of resonance captured by blue is here significant and appropriate. Whatever view we ultimately take on such issues (and, clearly, they can be addressed only in fine detail and in relation to particular examples), we should welcome Marmor’s analysis, for it enables us to raise the questions in a perspicacious form and to identify issues which might otherwise not have emerged with any clarity. The lesson to be learnt is perhaps one concerning the infinite depth, complexity and variety of those products of the human spirit that we crudely label ‘conventions’.