69
(2016) 30
journal for constitutional theory and philosophy of law
revus (2016) 30, 69–76
Dale Smith*
A New Type of Convention?
Some Remarks on Bruno Celano’s Pre-conventions
In “Pre-conventions: A fragment of the Background”, Bruno Celano argues for the exist-
ence and philosophical significance of what he calls “pre-conventions” – a type of con-
vention distinct from those hitherto discussed in the literature, and which transcends a 
number of orthodox philosophical distinctions. In these comments, I suggest that Cel-
ano may have shown that there is a distinct type of convention governing judgments of 
style or taste. If so, we may learn some important lessons (e.g. concerning the relation-
ship between conventions and rules) by examining this new type of convention. Howev-
er, I express doubts about whether the broader ambitions of Celano’s paper are fulfilled. 
I contend that pre-conventions are not as common as he suggests, and are not present in 
some of the other examples he gives. I also express doubts about whether he is right to 
suggest that there is a distinct type of convention that renders possible the development 
of conventions of the sort which David Lewis was interested in.
Keywords: Bruno Celano, pre-conventions, conventions, David Lewis
In a fascinating and wide-ranging paper, Bruno Celano argues for the ex-
istence and philosophical significance of what he calls “pre-conventions” – a 
type of convention distinct from those hitherto discussed in the literature, and 
which transcends a number of orthodox distinctions. Pre-conventions, says 
Celano, differ from both rules and de facto regularities, while possessing fea-
tures of each.1 Similarly, they transcend the fact/norm and nature/convention 
distinctions – they are “normative facts” which have features of both the natural 
and the conventional.2
The term “pre-convention” might be thought to describe something that is 
not itself a convention, but (in some sense) comes before genuine conventions. 
However, Celano is clear that this is not what he is claiming. Pre-conventions 
are, he says, genuine conventions.3 Indeed, he suggests that they are, in a sense, 
primary. As we shall see, he claims that they make possible the existence of cer-
* dale.smith@unimelb.edu.au | Melbourne Law School, University of Melbourne.
1 Celano 2016: 9.
2 Celano 2016: 9 and 30 (respectively).
3 If so, how can they transcend the nature/convention distinction, as Celano claims? I take 
it that the thought is that pre-conventions are genuine conventions, but possess features 
of the natural, and so help show that “nature” and “convention” are not mutually exclusive 
categories. See Celano 2016: 30.
70
(2016) 30
journal for constitutional theory and philosophy of law
Dale Smith
tain other conventions. More generally, while they typically remain unnoticed 
in the background, pre-conventions delimit the space in which other conven-
tions operate, by shaping our very activities and thoughts.4
Celano distinguishes pre-conventions from other types of convention on the 
basis that pre-conventions are not amenable to “rational explanation”, in the 
sense that there is no “‘train of reasoning’ explaining the behaviour of the rel-
evant parties.”5 In other words, pre-conventions do not involve “… the reasoned 
pursuit, by each of the parties involved, of their own goals”.6 This characterisa-
tion of pre-conventions is not entirely perspicuous, but it appears that Celano 
has in mind a contrast with views such as David Lewis’ claim that a convention 
is a way of satisfying people’s higher-order preferences in the face of a recurring 
coordination problem.7 Pre-conventions involve a shared way of doing things, 
but not one that can be traced back to an attempt by individuals to develop 
strategies to satisfy their higher-order preferences.
However, at times, Celano characterises pre-conventions somewhat differ-
ently. For example, he describes them as involving convergent behaviour “… 
which is not a biological regularity, and is the result of learning, but which is 
also automatic: it is spontaneous (unreflective), rapid, fluid, effortless.”8 He also 
refers to them as conventions that “… become as natural as breathing, a ‘sec-
ond nature’.”9 If this is intended to distinguish pre-conventions from Lewis-type 
conventions, then it strikes me as a mis-step. As Celano acknowledges, the fol-
lowing of a Lewis-type convention can also become “second nature”, and his 
attempt to show how this differs from the way in which pre-conventions are 
second nature returns us to the idea that the former, but not the latter, is sup-
ported by a “train of reasoning”.10
Before proceeding, I should say something about the ambitions of Celano’s 
article. At times, his aim appears to be relatively modest (e.g. to show that 
there is “space” for the existence of pre-conventions, and “that this space is not 
empty”11). However, at other times, he makes much bolder claims about the 
4 Celano 2016: 19.
5 Celano 2016: 11.
6 Celano 2016: 10. See also Celano 2016: 12 (referring to conventions that are amenable to 
rational explanation as “… the result of decisions by rational individuals pursuing their own 
interests, in the absence of an explicit agreement …”).
7 Lewis 1969. Celano refers to both Lewis and Hume (Celano 2016: 12) but, for simplicity, I will 
refer to these as “Lewis-type conventions”.
8 Celano 2016: 12 (footnote omitted, emphasis in original).
9 Celano 2016: 12.
10 Celano 2016: 14. At least, I take this to be Celano’s point. Read another way, he fails to 
distinguish pre-conventions from Lewis-type conventions, because he acknowledges that the 
latter, too, can be “tacit” in the sense in which he is interested.
11 Celano 2016: 19.
71Discussion 
(2016) 30
journal for constitutional theory and philosophy of law
philosophical significance of pre-conventions. Some of those claims concern 
the role that pre-conventions play in our lives. For example, he suggests that 
pre-conventions make the existence of Lewis-type conventions possible.12 In 
Section 5 of his article, he argues that induction requires judgments of salience 
that may be based on pre-conventions. And, in Section 6, he flags the possibility 
that pre-conventions underlie what he regards as the best account of concept 
acquisition and use.
Celano also suggests that the existence of pre-conventions calls into question 
several philosophical orthodoxies. I mentioned earlier that he thinks that their 
existence challenges some basic distinctions, such as the fact/norm and nature/
convention distinctions. And, if he is right about the existence and nature of 
pre-conventions, we must reject some widely held views about conventions – 
e.g. that they are necessarily a type of rule.13 As already noted, Celano claims 
that pre-conventions are distinct from both rules and mere de facto regularities.
I will not be able to explore all these issues here.14 Instead, I will focus on two 
of Celano’s central claims. The first is that pre-conventions are neither rules nor 
mere regularities of behaviour; the second is that pre-conventions are needed 
for the existence of Lewis-type conventions. In the course of examining these 
two claims, I will also consider whether pre-conventions do in fact exist.
To a significant extent, Celano elucidates the concept of a pre-convention via 
examples. However, many of these examples (e.g. soldiers marching or people 
playing chess) appear to involve Lewis-type conventions, since they involve the 
following of rules and are amenable to rational explanation. In the case of play-
ing chess, for example, Marmor suggests that there is a reason for the relevant 
conventions – namely, to satisfy our desire to play a competitive, intellectual 
board game.15 Those conventions also take the form of rules (the rules of chess).
12 Celano 2016: 22.
13 See, e.g., Marmor 2009: 1–2. Marmor interprets Lewis as denying that all conventions are 
rules (at 15–16), but perhaps the better reading is that Lewis claims that some conventions 
are penumbral, rather than central, instances of rules: see Lewis 1969: 104–105. I think that 
Celano would go further, and deny that some conventions are even penumbral instances of 
rules.
14 In particular, I cannot discuss properly Celano’s suggestion that pre-conventions underlie 
the best account of concept acquisition and use (which he takes to be an internalist, 
Wittgensteinian account). However, let me very briefly note that, if we instead endorse an 
externalist account (and I have expressed sympathy for such an account elsewhere: Smith 
2012: 90–95), then the role, if any, that pre-conventions play in this context will be much more 
limited.
15 Marmor 2009: 41. Marmor claims that the relevant conventions are “constitutive” conventions, 
not coordination conventions of the sort Lewis was interested in. (See Marmor 2009: 31 
regarding this distinction.) I take it, though, that pre-conventions are to be contrasted with 
both coordination and constitutive conventions.
72
(2016) 30
journal for constitutional theory and philosophy of law
Dale Smith
Celano seeks to challenge this thought by pointing out that the activities 
in question can be internalised, in the sense of being performed automatically 
or unreflectively or as a matter of second nature. Indeed, they may be a matter 
of know-how – someone may know how to march, or play chess well, without 
being able to express that knowledge in propositional form – and, Celano con-
tends, know-how is not reducible to a set of rules.16
However, care needs to be taken in analysing these examples. It is plausible 
that playing chess well involves both abiding by certain conventions and a cer-
tain know-how that cannot be reduced to a set of rules. However, it does not 
follow that the conventions can be equated with that know-how, so as to entitle 
us to say that the conventions are not reducible to a set of rules. Indeed, it is 
tempting to say that the conventions of chess are (by and large) confined to the 
rules of chess, and that the know-how involved in being able to play chess well 
is non-conventional. For one thing, that know-how need not involve any form 
of agreement, even in Celano’s extended sense of “agreement”.17 We can imagine 
a community in which only one person possesses that know-how; everyone else 
plays chess poorly.
That said, another of Celano’s examples – concerning judgments of style or 
taste – is less easily dismissed. If we leave aside taste in, say, ice cream, and fo-
cus instead on, say, clothing or art, then it is plausible to suggest that there is a 
conventional aspect to judgments of style or taste. It is also plausible to suggest 
that these judgments are, in part, matters of know-how, rather than knowing-
that. As Celano points out, attempts to reduce matters of style or taste to a set of 
propositions, such as one might find in a handbook or how-to guide, are typi-
cally inadequate and may themselves display a lack of style or taste.18 Here, we 
seem to have a subject-matter that is governed (in part) by conventions and yet 
cannot adequately be reduced to a set of rules. Moreover, it is more difficult than 
in the case of chess to separate the conventional aspect of the practice of making 
judgments of style or taste from the aspect of the practice that is irreducible to 
rules. Indeed, it may be part of the relevant conventions that one must possess 
a certain attitude, of not trying to reduce judgments of style or taste to a set of 
rules to be followed. (This may itself amount to a rule, but it entails that other 
conventions governing judgments of style or taste are not reducible to rules.)
I put the point tentatively, because there are several objections that would 
need to be addressed before we could confidently conclude that judgments of 
style or taste are governed by conventions that are irreducible to rules. For ex-
ample, it would need to be explained how a breach of the conventions govern-
16 Celano 2016: 16–17.
17 Celano claims that all conventions involve agreement, but he uses the word “agreement” very 
broadly: Celano 2016: 5.
18 Celano 2016: 17.
73Discussion 
(2016) 30
journal for constitutional theory and philosophy of law
ing judgments of style or taste warrants, or is taken to warrant, criticism if those 
conventions are not rules.19 Nevertheless, if these objections can be met, then 
Celano has succeeded in identifying a type of convention that is not reducible to 
rules, and so differs importantly from both coordination and constitutive con-
ventions. This would represent a significant contribution to our understanding 
of the nature of conventions.
However, it would not suffice to fulfil Celano’s broader ambitions. I have sug-
gested that the phenomenon he has identified is less common than he claims. It 
does not extend to other of his examples, such as playing chess or marching.20 
Nor have we yet been given any reason to think that this new type of convention 
is necessary for the existence of Lewis-type conventions or that, more generally, 
it shapes the boundaries of our activities and thought.
To obtain a better sense of whether these broader ambitions can be fulfilled, 
let us consider Celano’s discussion of the relationship between pre-conventions 
and Lewis-type conventions. He points out that, to follow a Lewis-type con-
vention, one must identify both the action that is required and the situations 
in which that action is required.21 Determining whether a convention applies 
to a particular situation typically requires one to ascertain whether that situ-
ation is relevantly similar to past situations in which the convention applied 
(what Lewis calls “precedents”). However, as Lewis points out, no two situations 
are exactly the same, and so there are always multiple ways in which we could 
draw analogies or dis-analogies between the present situation and a precedent. 
Nevertheless, we tend to draw analogies in more or less the same way, and this 
enables precedents to guide our behaviour. As Lewis puts it, “[w]ere it not that 
we happen uniformly to notice some analogies and ignore others /…/ prec-
edents would always be completely ambiguous and worthless.”22
Celano suggests that our tendency to all draw the same analogy is often due 
to the existence of a pre-convention. While sometimes the drawing of the same 
19 See Marmor 2009: 16 for a similar objection, albeit in a different context.
20 An anonymous referee suggested that, once it is conceded that there may be a distinctive type 
of convention governing judgments of style or taste, we must revisit the suggestion that there 
is no room for pre-conventions when it comes to playing chess.  Consider card games such as 
poker, which appear (for Celano’s purposes) to be relevantly similar to chess: we might think 
that professional poker players share a certain style that is both conventional and irreducible 
to rules.  I do not know whether this is true, but – if such conventions exist – they would 
appear to be separable from playing the game itself (or even from playing the game well: as 
noted earlier, the know-how involved in playing the game well could conceivably be possessed 
by only one person).  If so, their presence is, at most, a contingent feature of some instances of 
playing a card game, or chess.  This might, however, suggest that judgments of style or taste, 
and the distinctive type of convention that may underlie those judgments, are more prevalent 
than first appears.
21 Celano 2016: 20–21.
22 Lewis 1969: 37–38 (quoted in Celano 2016: 21).
74
(2016) 30
journal for constitutional theory and philosophy of law
Dale Smith
analogy may be hardwired into all human beings, in other cases it depends on 
“… more or less arbitrary, agreements  /…/ which are neither mere rules nor 
mere regularities, but which partake of both – they guide action, fixing the cor-
rect way to proceed: pre-conventions.”23 These pre-conventions facilitate the 
regularity of behaviour among members of the relevant group that is necessary 
for a Lewis-type convention to exist, and hence render Lewis-type conventions 
possible.
It is important to recognise that the issue here concerns how Lewis-type 
conventions are established, not how they are maintained once they are estab-
lished. Consider Lewis’s example of a convention governing who should call 
back when a phone call is cut off. As he points out, once we have enough prec-
edents (i.e. past situations where this coordination problem has been solved), 
we can identify a rule governing the situation.24 However, initially, where we 
have only a limited number of precedents, no unique rule can be identified. 
For example, perhaps A and B have faced this situation only once before, and 
that was a case where A placed the initial call. In that case, it was A who rang 
back after the call was cut off. Now imagine that, this time, it was B who called 
A. Should B call A back (on the basis that, in the earlier precedent, it was the 
person who placed the initial call who rang the other person back) or should A 
call B (on the basis that it was A who called back last time)? Lewis suggests that 
it just happens to be the case that we tend to answer this question in the same 
way (in this example, we draw the former analogy, not the latter).25 Celano can 
be understood as suggesting that this convergence amounts to a pre-convention 
– not a rule, but something more than a mere regularity of behaviour – which 
makes the development of the Lewis-type convention possible. However, for 
the reason Lewis gives, if pre-conventions are needed, it is only at the formative 
stage of the development of the Lewis-type convention.26
Of course, if pre-conventions are needed at any stage in the development of 
a Lewis-type convention, then identifying this fact is a major contribution of 
Celano’s paper. However, are they needed? It might be suggested that positing 
the existence of pre-conventions helps make sense of how Lewis-type conven-
tions come about.27 In particular, one might think, it advances our understand-
23 Celano 2016: 22 (emphasis in original).
24 Lewis 1969: 39–40.
25 Lewis 1969: 37–38.
26 Celano might contest this on the basis that, even at a later stage, identifying the relevant rule, 
and following it, depends on pre-conventions. This would be part of a broader view that rule-
following in general requires pre-conventions. I cannot assess that view here, beyond noting 
that Celano is right to suggest that it stands or falls with his views about concept acquisition 
and use (concerning which, see above fn. 14).
27 Though cf. Marmor 2009: 21 (suggesting that a philosophical account of conventions should 
not speculate about how conventions, as a matter of historical fact, emerge).
75Discussion 
(2016) 30
journal for constitutional theory and philosophy of law
ing of this process further than the suggestion that it just happens to be the case 
that we tend to draw the same analogies.
I am not sure, however, that positing the existence of pre-conventions ad-
vances our understanding all that far. In this context, a pre-convention amounts 
to a tacit agreement about how to proceed, in circumstances where it was open to 
us to proceed differently (by drawing different analogies between situations).28 
Given the broad sense of “agreement” used by Celano, positing such an agree-
ment may not take us very far beyond the suggestion that we all just happen to 
draw the same analogies in the early stages of developing a Lewis-type conven-
tion.
Celano might insist that the key point is that the tacit agreement underly-
ing the development of a Lewis-type convention is not amenable to rational 
explanation. (Recall that this is meant to be the hallmark of a pre-convention). 
However, it is not clear that this is true. If achieving convergence on which situ-
ations are analogous is necessary for a Lewis-type convention to become estab-
lished, then achieving that convergence is clearly in the interests of the relevant 
parties.29 Celano might respond that the parties are not consciously pursuing 
their interests, but rather acting automatically, as a matter of “second nature”. 
However, as noted above, it is doubtful whether this is sufficient to distinguish 
pre-conventions from Lewis-type conventions.
Where does this leave us? Celano may have shown that there is a distinct 
type of convention governing judgments of style or taste. If so, we can learn 
some important lessons (e.g. concerning the relationship between conventions 
and rules) by examining this new type of convention. However, there is reason 
to query whether many of Celano’s broader ambitions are fulfilled. It seems that 
pre-conventions are not as common as he suggests. I have contended that they 
are not present in his examples of marching or playing chess. And I have ex-
pressed doubts about whether there are distinct, “pre-” conventions that render 
possible the development of Lewis-type conventions.
—Acknowledgment.—  Thanks to Lulu Weis for very helpful comments on an earlier 
draft, and to an anonymous referee for Revus.
28 This reflects Celano’s suggestion that conventions involve an agreement and a certain type of 
arbitrariness (Celano 2016: 10), and that pre-conventions are a type of tacit convention (one 
which is “… not backed by a train of reasoning”: Celano 2016: 12).
29 See Sugden 1998: 404 for the suggestion that this is just another coordination problem to be 
solved.
76
(2016) 30
journal for constitutional theory and philosophy of law
Dale Smith
References
Bruno CELANO, 2016: Pre-conventions. A frag-
ment of the Background. Revus – Journal for 
Constitutional Theory and Philosophy of Law 
(2016) 30: 9–32. Available at: http://revus.revues.
org/3386.
David K. LEWIS, 1969: Convention: A Philosophical 
Study. Cambridge (Ma.): Harvard University 
Press.
Andrei MARMOR, 2009: Social Conventions. From 
Language to Law. Princeton/Oxford: Princeton 
University Press.
Dale SMITH, 2012: Must the Law Be Capable of 
Possessing Authority? Legal Theory 18 (2012): 
90–95.
Robert SUGDEN, 1998: The Role of Inductive 
Reasoning in the Evolution of Conventions. Law 
and Philosophy 17 (1998).