Rod Watson's Constitutive Practices and Garfinkel's Notion of Trust: Revisted (2009)

Journal of Classical Sociology
Copyright © The Author(s) 2009. Reprints and permissions: Vol 9(4): 475–499 DOI: 10.1177/1468795X09344453
Constitutive Practices and Garfinkel’s Notion of Trust: Revisited

ROD WATSON LIAS, Institut Marcel Mauss, Paris
ABSTRACT This article is intended to reinstate, in at least a prefatory way, some ethnomethodological (EM) considerations concerning trust. The idea of consti- tutive practices – as it was taken up in Garfinkel’s sociology – turned on trust as a background condition for mutually intelligible action. Starting with a consideration of Garfinkel’s 1963 study of trust, the article critically considers some formal analytic alternates to his approach. The aspects of trust that are ‘elusive’ to the formal-analytic approach are shown to result from its allusive treatment by formal analysis. In Garfinkel’s hands trust is not elusive. The critique of formal analytic studies builds on Garfinkel’s writings and certain strands of analytic and ordinary language philosophy. These sources ground the author’s suggestion that the study of trust be taken up again, albeit along respecified analytic lines. Examples are given, both of an EM and conversation-analytic (CA) kind.
KEYWORDS conversation analysis, ethnomethodology, formal analysis, game theory, membership categorization, ordinary language philosophy, rational choice, respecification, rules, trust
In the overlap between philosophy and sociology found at the intersection of John Rawls’ ‘Two Concepts of Rules’ (1955) and those who have taken up the various approaches that constitute ethnomethodology (EM), the idea of ‘Trust’ plays an important role. One of the basic conditions of any constitutive practice is a mutual commitment to rules of engagement in that practice – that is, all parties to the interaction must understand that they are engaged in the same practice, must be competent to perform the practice, must actually perform competently and assume this also of the others. It is as such a constitutive condition that
Garfinkel (1963) elaborated trust as a necessary background condition of any mutually intelligible interaction.
While the importance of trust as a social phenomenon is generally recogn- ized and the issue of trust has been around in the background in social science since at least the 1950s, few approaches to trust consider it in the context of constitutive practices – treating it rather as just one more phenomenon to be elaborated within a formal-analytic approach. In this paper, I shall point to a few of the most salient early studies that are not EM and very briefly indicate their formal analytic properties. After an initial critique of these I will then go on to consider Garfinkel’s analysis of trust, one of the earliest, yet somewhat neglected. I will recommend that his article on trust (1963), and the phenomenon of trust, be subjected to a reconsideration.
Trust and Sociology
Half a century ago, Pitirim A. Sorokin wrote a coruscating critique of the sociology of the time, Fads and Foibles in Modern Sociology (1956), and while EM has taken another route, formal analytic sociologies still seem driven by such fashion and fetish. Since then, ‘trust’ seems to be one of the things that has increasingly come to be fetishized. ‘Big hitters’ – Anthony Giddens, Niklas Luhmann, and so on – have written about it, and from being a shadowy presence in sociology, trust has become a highly visible one. Indeed, in the hyperbolic tropes so common in soci- ology, we hear much about a ‘crisis of trust’. Yet again, this exhibits mainstream sociologists’ attachment to the apocalyptic. However, these claims all seem to me to beg some fundamental questions.
Given the early status of Garfinkel’s study and the fact that he never explicitly returned to issues of trust, I shall then make some still tentative and provisional suggestions about how ‘trust’ might be approached according to Garfinkel’s most recent updating of EM and (so far as this short article allows), according to some analytic-philosophical concerns with ‘logical grammar’. If one endorses the notion that the early work represents a ‘proto-EM’, then the trust paper might be seen as standing ‘on the cusp’ between that and ‘full-on’ EM. But some of it was, after all, re-presented in Studies in Ethnomethodology (1967).
One reason for considering Garfinkel’s paper is that it has frequently been grafted onto formal analytic studies – Giddens’ (1990, 1991) work gives us one example, as does Barbara A. Misztal’s (1996) – in ways that to varying extents occlude the implications of this early ethnomethodological study. In particular, the radical methodological implications are, to varying degrees (for example, in Giddens more than Misztal), smothered. Instead of treating trust as a tacit and necessary precondition, it is rendered as attitudes, and so on, that can exist more or less in persons and situations – losing its constitutive relevance. As such, Garfinkel’s study of trust is placed under house arrest. I hope to suggest how to remove such confining considerations.
In much contemporary formal analytic sociology, issues of trust or mistrust are conceived as omnirelevant – again, Giddens’ notion of trust in abstract systems, or trust in urban public spaces, in politicians, and so on. However, it must be said that trust as a first-order phenomenon has proved remarkably elusive. Part of the reason for this is that, because of its tacit character, trust is such a fugitive contextual phenomenon – it ‘appears’ in only a few contexts and ‘disappears’ in very many others. Most frequently, it is a presumptive phenomenon and therefore tacitly attended to by members.
This elusiveness has been compounded by the allusiveness of the standard formal analytic technologies of social research. These technologies have signally failed to capture that first-order phenomenon, and particularly its contextual nature (perhaps some of the ‘Chicago School’ studies – Marvin B. Scott, Robert E.L. Faris, and so on – are possible but partial and elliptical exceptions to this). Indeed, their approach to the issue of trust shows the ways in which the very design features of these research technologies lead them to change the subject, to present trust in terms of something else – trust as choice, as attitude, as risk, as a game, and so on. Or trust is elided through the conjunction ‘and’ – ‘trust and the family’, ‘trust and abstract political systems’, and so on, where what we actually get is a study of family organization, the nature of abstract systems, and so on. In such cases, trust tends to be ‘defined’ by its object.
Or trust is treated purely aprioristically, as captive within the confines of a particular theory – ‘trust as habitus’ in the Pierre Bourdieu repertoire, trust as function in Luhmann’s framework; indeed, Luhmann claims that trust can only be known in its function, as reducing social complexity, and so on (Luhmann, 1979: 93). At best, in formal analysis, trust is addressed by indirection, and the indirection is brought about through the conjunctive or adverbial reductions of the actual phenomenon.
Even where the contextually variable aspects of trust are noted by formal analysis, this type of analysis lacks the technology to render that variability. It is after all a technology for reducing variability to patterns. For instance, Luhmann acknowledges such circumstantial variability, but it perforce remains for him largely a noted thing – perhaps a ‘problem’ – rather than one that re-casts his analysis. If one is committed, as Luhmann is, to theoretically stipulating social order in the abstract – in Luhmann’s case a modified functionalism – then the ‘chiaroscuro’ of trust will perforce be lost.
Thus, the apparent elusiveness of trust is compounded by its allusive treat- ments in the formal analytic corpus. One upshot of all this is that these treatments take on an intersubjectively problematic status, that is, their relation to trust as a phenomenon of ordinary membership is, at best, indeterminate or conflated and at worst the phenomenon is subjected to methodologically ironic reduction or relativization. The array of approaches in rational choice theory, decision theory and game theory shows this intersubjectively problematic status. For instance, such indeterminate status is found in the rational choice theory of Diego Gambetta
(see, for example, Gambetta, 1988), who invites us to consider the conditions under which trust is justified, where trust or mistrust are invariably treated as ‘outcomes’ of these conditions. Another such example is that propounded by James S. Coleman (1990).
In a sense, we might see Coleman’s analysis as coming as close to identi- fying trust-as-such as is possible under the formal analytic rubric. Essentially, Coleman had the aim of at least prototypically quantifying trust. He conceives of trust as a calculated risk taken about the performance of others, given the time-lag before that performance is actually issued.
Coleman formalizes the conditions under which ego will invest trust in alter, and his formula includes factors such as the likelihood that (according to a standard estimate) alter is trustworthy (p), or that he is not (1 – p). L is the loss should alter prove to be untrustworthy and G the gain if he is. He thus devises a formula for when ego will trust (or distrust) alter. The formula is p + 1 – p < L + G. This formula expresses, for instance, that ego will invest trust if the likelihood is that alter is trustworthy or not is greater than the gain or loss respectively. Much, Coleman says, depends on the level of information available, though (significantly) that is not built into the formula.
The extent to which such judgments would themselves be based on infor- mation only available to ego and alter when they are mutually engaged in and committed to a constitutive practice is not considered. Garfinkel (2008[1952]) would of course argue that there can be no information in the first place without trust. Some practices – constitutive practices – require as a necessary background condition that ego trust not only alter, but all participants in the practice in question. Trust in this sense does not mean to trust the whole person in all of their aspects, but, rather, to trust only that they are committed to this practice, competent to perform it, and that they trust this of you. Not to trust in this way is to fail to participate (or have information) altogether.
Other variants of Coleman’s approach involve game-theoretic models where trust or mistrust are treated as outcomes of a set of calculated game moves and conditions, again based on a given level of information. Once more, we see trust as ‘revealed’ or formulated by the application of a general, professional- technical methodology of (at least an ersatz) mathematical kind, and trust as an object is conceived generally and treated abstractly. Of course, there is no study of how an assessment of likelihood, and so on, is arrived at as a sensible phen- omenon: such issues of sense-making are completely elided.
From the EM standpoint, one might well consider the formula as quite blatantly yielding a reductive, attenuated view of trust as a phenomenon of order – or what Howard Schwartz and Jerry Jacobs (1979) dubbed an ‘anemic’ char- acterization, with the phenomenal detail leeched out. However, one must say that Coleman’s formula meets many of the ideals and criteria of formal analysis: at least it tries to focalize trust per se and to do so in a formal, generalizable way, one that allows of at least an ‘apparent’ mathematization, although, as ever in social
science, this apparent mathematization still retains the character of a promissory note rather than that of an actually delivered promise – certainly so far as trust is concerned.
Whilst we might say that Coleman’s formula is an idealization, that is just what formal analysis is about: if it aims to deal with a particular case at all, it tries to do so by extracting what are purported to be its purely interchangeable features, the features that might make it formally similar to (all) other cases irrespective of the occasioning and formative context, and so on, and irrespective of lay members’ own reasoning on the matter. Trust is thus treated in such analyses as a unitary phenomenon, reproducible, transposable and duplicative. This does not, of course, render the formal analytic conception immune to criticism, but it does mean we should be careful about the criticisms we do adduce. A prior move, anyway, is to revisit the EM conception of trust and Garfinkel’s early study.
The idea of constitutive features – such as trust – as actually prior to and constitutive of action and objects (within a practice) rather than as emerging from – as outcomes of – action stands as the essential difference between what Rawls (1955) called summary rule orders and constitutive orders. Garfinkel takes up trust as a necessary background condition in a constitutive order – not a summary rule order – and this distinguishes his approach from most of sociology, which takes a formal analytic and summary rule approach.
Part of Garfinkel’s study involved a very well-known set of what have been called ‘breaching experiments’ but which he later referred to as ‘tutorials’ and which I shall term ‘interventions’ – interventions, that is, into the normal stream of daily life as members experience it. Some interventions were game-based, such as getting a student playing noughts and crosses (or tic-tac-toe) to write, say, a nought on a line rather than in a box, or by making another move not provided for in the game’s rubric. However, these interventions eventually ceded to non game-based situations, such as getting a student to remorselessly pursue the elaboration of a point in a conversation whilst presenting that as a perfectly ordinary conversational activity: for example, when a co-conversationalist reported a flat tire, the student would pursue that through asking ‘What do you mean, a flat tire?’; or, in another instance, to commence bargaining in a shop with fixed prices. We might see these interventions as attempts to raise into visibility matters that are typically taken for granted, matters (such as trust and constitutive orders) that can then be turned into topics for inspection on their own behalf – somewhat akin to a phenomenological bracketing exercise.
There is nothing ludic or mischievous about these interventions. There is a profoundly serious methodological reason for them, namely to focalize trust- as-such (and the constitutive practices that require it) rather than addressing it through formal analytic allusion, or by treating it as a residue of, or reduction to, something else (e.g. ‘untrustworthy identities’). Misunderstood by those taking a formal-analytic (summary rule) view, Garfinkel’s interventions came to be notori- ous in social science, seen either as frivolous, maliciously ludic or, somehow,
ethically dubious: these accusations often came from social scientists not previously known for their Jane Austen-like delicacy. In fact, Garfinkel’s interventions were serious teaching exercises quite as much as research ones, and in this respect his article on trust presages his later work on ‘teachability’ as a research device (see also below on Harvey Sacks’ methodological requirements).
Rules and Practices
What was revealed through these interventions? For Garfinkel, at that time, trust had its ‘natural home’ amongst fundamental considerations such as the nature and status of rules, sense-making practices, and the like – that is, in the ordered properties of the achieved intersubjectivities of everyday life. In particular, he argued that basic rules with their ‘constitutive accent’, or reciprocally endorsed constitutive expectancies, turn out to be central to the phenomenon of trust; this in contradistinction to the more discretionary rules of normative preference. Basic rules yield a sensible order – and as such breaching them threatens the sense of things. By contrast, if a rule of merely normative preference is breached, the player might well be seen by fellow players as deviating within a sensible order without threatening it as a set of perceivedly normal game events. In addition, we have game-furnished conditions which provide for how the game is played according to the constitutive rules and constitutive structure of the game.
Let us take, for instance, a game of mixed doubles in tennis (I apologize to game theorists for referring to a real game). A basic rule is that, say, a player is permitted to hit the ball hard and directly at the body of an opposing player. However, a preference rule – in the main, only activated in club play – is that a male player should not hit the ball directly at a female opponent. If he does, the umpire will not rule against that shot, but yet it deviates from the rules of preferred play.
In everyday life, Garfinkel says, the relatively formalized and precise rules of a game find a rough and ready equivalent – and no more than that – in the pre- suppositions comprising the attitude of daily life: members’ natural attitude. Trust comprises participants’ reciprocal endorsement of these presupposed matters and their consequent maintenance of sensible social order. In games, trust comprises players’ reciprocal orientations to basic game rules. Whilst games do not really carry over to ‘ordinary’ or ‘serious’ life, the game model does serve to clarify dif- ferent orders of rule and how rule-using players can presage the actions of others relative to their own.
John Heritage (1978: 93–5) provides a subtle characterization of the work done through rules in relation to sense-making practices. Broadly put, these practices supply the situated sense of the rules but, reflexively, rules operate to furnish an instructed gloss or formulation of the practices. As Heritage puts it (1978: 93), arrays of practices in social settings (and the descriptive particulars
constituted by these practices) ‘collect’ around rules, and rules instruct users as to setting-appropriate conduct within a specific setting.
Parenthetically, by conceiving of the working of rules in this ‘procedural’ way, Heritage notes that there is a basis of complementarity between EM and CA: the former focalizes the work of methodically establishing a sensible order in settings, and the latter explicates the sheer systematicity in the ordering of the procedures themselves. He thus, incidentally, furnishes a basis upon which at least some of those who wish to do so may initiate analytic moves to (re-establish) such a continuity and to head off those attempting to establish and reify a divergence between EM and CA.
Garfinkel’s distinction between basic and normative rules clearly addresses these selfsame issues. Classical sociologies tend to focus exclusively on the latter, conceiving of conduct as more or less approximating to a regulative norm, to use John Rawls’ terms, where beyond certain limits, conduct violates expectations and requirements and is deemed deviant. Societies are often seen as built on the basis of such normative rules and are thus analytically presented as systems of regulated social action.
Garfinkel, however, poses a question of priority: he asks how is the action commonsensically identified as whatever it is in the first place? This is a constitutive order question. In order for the moral-regulative ‘machinery’ to be applied to a given action (or set of collective actions), the action has to be understandable to parties in the first place. Here, Garfinkel’s conception of constitutive rules or arrays of constitutive or basic practice becomes relevant.
Basic rules might be conceived as glosses of arrays of constitutive practice that comprise a local gestalt contexture. They are sense-making instruments deployed in situ: known and used in common. This means there is a reciprocity in the use of these rules. It is assumed by parties to a setting that their co-participants see those circumstances, mutatis mutandis, in similar terms and that this would hold were those parties to exchange places.
This, in turn, gives us the most primitive sense in which ‘trust’ figures in social life. As parties to a given contexture, we place trust in that reciprocity and interchangeability. We trust in other parties’ ability and motivation to make similar sense of a situation, using similar sense-making methods and instruments.
Of course, there is always a risk of reifying even the most analytically valuable distinction, and so it is with the ‘constitutive’ (basic) – preferential (normative) – one. A logical distinction and logical prioritizing may yield complications in empirical analysis. In this respect, Heritage’s formulation, above, is particularly useful in that it brings together the constitutive and regulative aspects of rule use: the constitutive and regulative appresent themselves as two facets of the use of a given rule, as does the notion of rules as instructions.
However, Garfinkel substantially rescinds the game model in his character- ization of everyday life for the following reasons. Games comprise what he terms
‘encapsulated episodes’: in terms of their conducting consequence, they are sealed off and relatively impervious to external contexts. Game rules, and differences of status as between rules, are less well defined in everyday life, and games, of course, typically comprise bounded time-out from ‘serious life’. As Garfinkel’s ethnomethodological work progressed through the years, the emphasis on ‘rules’ as such continued to diminish – to be replaced by ‘instructions’, ‘instructed action’, and the like. Even as his 1963 article proceeds, the emphasis on rules lessens quite notably.
In the settings which are the natural loci of the attitude of everyday life, we must instead refer to the procedural character of perceivedly normal environments, to their constitutive order – and, what will be pertinent to this paper, everyday life events tend to be less episodic, less calculable, more open- textured, more permeable than games with their bounded rule-sets and conse- quent sealed-off character. The projectable consequentiality of moves in everyday settings is less amenable to specification. As opposed to game events, everyday life events – including those characterized by substantive conflict1 – involve presupposed typifications, precedents, morally sanctioned schemata, and so on. These themselves are, one might say, trusted resources in the sense-making process. Moreover, as Garfinkel’s study of an intersexed person shows (‘Agnes’, Garfinkel, 1967), people do not live their everyday lives in episodic ways (pace Erving Goffman). Rather, in their use of experiential precedents and the like, they build their past seamlessly into their present understandings of their conduct; this is what the durée is all about.
In everyday life situations, trust, then, comprises parties’ reciprocal adherence to the sensible order of events thus produced. It involves mutual con- formity to the ‘constitutive accent’ of these presupposed elements, not least in the face of equivocality or open-endedness. It also involves the interchangeable nature of participants’ orientations to these perceivedly normal environments. In this basic respect, trust is, analytically speaking, a condition for the stabilization of these normal environments, as Garfinkel indicates in the very title of his article, ‘... “Trust” as a Condition for Stable Concerted Actions’ (1963). When the trust condition is not in place, participants experience bewilderment, confusion, frustration or indignation, or they attempt to make sense of or normalize the events in different terms – as a joke, or hoax, a deliberate provocation, obtuseness or whatever. In this view, trust operates as what Alfred Schutz (1967) terms a ‘scheme of interpretation and expression’ or what we might term a ‘background scheme’. There are some resonances, however distant, to Simmel’s notion of trust as a background condition of exchange (Simmel, 1950: 313–36; 1978: 178–9), or Émile Durkheim’s notion of the non-contractual elements of the contract (1933 [1893]: Book III).
Durkheim posed an argument against contract theorists of a more or less individualist persuasion – utilitarians and those influenced by utilitarian individualism such as Herbert Spencer. This line of thought held that contracts
brought into alignment and reconciled the individual interests of parties, to their mutual benefit in a given situation. It also held that contracts therefore form a basis of social order.
Durkheim’s response is that for contracts to work at all, a set of binding expectations must pre-exist any particular interested contract with all its specific contingencies. These expectations must transcend the contingencies of any given contract and the various interests of each individual party. The rules are conse- quently not merely legal but also moral and constitute the ‘background’ informing entry into any given contract. Such expectations control the recognition of a contract, the conditions under which contracts may be understood as binding and their consequentiality. So, prior tacit and constitutive ‘contract’ makes regulative contract possible.
Just as constitutive rules comprise a condition of the formation, management and enforcement of contracts, so trust might be seen as a normal condition that informs parties’ entry into any given interaction. It constitutes part of anyone’s basic understandings of the local order of that interaction and involves the pre- sumption that the other parties will orientate themselves to the interaction in similar, interchangeable ways. This, then, is one of the ways in which trust might be seen to be a ‘scheme of interpretation and expression’ whereby it is prior in the sense of ‘presumptive’ and yet specifically established, fine-tuned, and so on, in situ, for ‘another first time’. The intersubjective architecture (to modify Heritage’s phrase) of any given interaction thus ‘expresses’ trust and is oriented by it.
If trust itself is a constituent element of any specific, local interaction, the methods through which it is ascribed or invested might be seen, with all due attention to differences in situated detail, as presumptive ‘conditions’ for that interaction. Of course, a central feature of the trust that is so invested involves a presumption that the expectations that Durkheim calls ‘non-contractual elements’ will be methodically, reciprocally and locally known and administered as a set of relevancies in the situation.
We might say, then, that in Garfinkel’s initial formulation, trust operates within the texture of constitutive expectancies of daily life even though it may not be straightforwardly reducible to such expectancies. These expectancies, glossed by the term ‘constitutive accent’, refer to the expectation that rules be followed irrespective of participants’ particular motives, wishes or circumstances, that the rules are to be adhered to by all parties to a given setting.
Garfinkel’s distinction between basic and preference rules falls into what was at the time a recent history of broadly parallel distinctions, including John Rawls’ distinction between constitutive and regulative rules. The former serve to define a given activity as what it is, while the latter control prior, already-defined actions. Garfinkel’s preference rules broadly belong to that regulative (or summary rule) category, whereas his basic rules are of the constitutive type.
In many of its incarnations, trust is part of the way in which members realize their local environments as perceivedly normal environments of familiar objects,
identities and secured activities or situations. Part of this involves members’ reciprocal orientation to the normative accountability of their actions, where by ‘reciprocal’ I mean that a participant not only holds her/himself to a given orientation but takes it that co-participants will hold to it, too. All this is part of the practical logic of constitutive order as it is built into and expressed in naturally occurring conduct. It is thus ‘taken on trust’ that all parties to a given setting are using the same presuppositions and methods to identify it and act within it. Trust ‘underwrites’ that interchangeability, as it were, and as such is also integral to the morally binding nature of common understandings.
Trust, Games and Rational Action Theory
What, then, can Garfinkel’s formulation tell us about rational choice and game- theoretic models of trust? We must again note that in Garfinkel’s article the game model, for all its drawbacks, operates in a manner somewhat akin to a per- spective by incongruity, a planned misnomer designed to cast new light on an old phenomenon.
Of course, criticism of formal-analytic sociologies may also be seen as a problematic use of EM. However, I consider that if proponents of sociologies are employing whatever resources to make what they purport to be analytic claims, then we can consider them ‘fair game’, not least because such an exercise can help us further define the alternate status of the two types of approach.
In relatively recent contemporary formal analytic work such as that of Coleman, Giddens and Adam Seligman (2001), trust is frequently, even typically, conceived in association with ‘risk’. Again, each of these conceptions shares an inter- subjectively problematic status, that is, we are never sure whether the concept ‘risk’ in these approaches is one that is purely stipulated by the analyst and driven purely corroboratively through a set of empirical materials, or whether it is grounded in members’ own understanding of their situation, or, indeed, whether it is an indeterminate conflation of the two. Students of modernity and postmodernity often decry nineteenth-century evolutionism with its postulates of monocausality, unilinearity, and so on, but these selfsame students have no compunction in characterizing contemporary society according to a single descriptive feature such as ‘risk’. Such a descriptive technique is not, then, restricted to nineteenth-century evolutionism: de facto, single-factor analysis remains part of the armory of formal analytic sociology.
In a way, linking trust with risk might be a variant of a very frequently found technique, that is, studying trust through its negation. However, if we take Garfinkel’s view of trust, we treat trust as a presumptive element in all concerted action irrespective of its ‘risky’ character. Following Garfinkel, we could treat trust or distrust in a risk situation as a special determination of those matters, but not as itself comprising the generic nature of trust.
Here we need to digress for a moment on the issue of method. By and large my approach follows Peter Winch (1958). I argue that philosophical ‘technique’ can be usefully employed to resolve particular linguistic/conceptual confusions. Indeed, it seems to me that the specification of the nature of confusions about the general character of language might be held to inform and provide for the tasks of elucidation and clarification of specific concepts and their implications, the establishing of conceptual conjunction and disjunctions, and so on. To these tasks I should add another, that of removing false or artificial restriction in the use of ‘following through’ of a given concept, and it is to this issue (inter alia) that I now turn.
We might extend our argument on trust as a condition of concerted actions to the issue of trust as being conceived by most scholars as invariantly an outcome. This is just the kind of artificial restriction to which I am referring. Of course, we might not deny that, on occasion, something we want to call trust can also be an outcome – and an outcome of calculation, ratiocination, and so on. However, whether rational choice models with their idealizations of rationality, of games or of decision-making models actually model the logic-in-use leading to such occasioned outcomes is dubious in the extreme. What we are saying is that trust is not invariably an outcome. Nor does the possibility that trust can be an outcome preclude the requirement that in that same case it can also be a condition, for example that trusted resources are necessarily employed in the pro- duction of that outcome.
The formula adduced by Coleman and others depends largely upon the concept ‘information’. But ‘information’ as they conceive of it is a classic formal analytic and, indeed, highly problematic abstract gloss of a diverse range of matters pertaining to members’ – what they encounter as a ready-made, usually taken for granted (trusted, one might say) – world of normal social types, types of action, kinds of context, and so on. Members certainly do not invariably conceive of their world in terms of a carefully or exhaustively set-out range of alternatives cast in terms of ‘information’ (let alone ‘perfect information’) and make decisions amongst them (for comments on the how the term ‘information’ has been used, see Garfinkel, 2008 [1952]). To invariably start from a situation of no trust seems to be akin to a doctrine of general skepticism. But for members to somehow decide in each case to invest trust would in fact be a recipe for inaction.
We might add that the diverse arrays of typifications, and so on, are all organized according to a diverse set of forms of life – religion, art, medicine ... – each of which is composed as, again, diverse arrays of knowledge. We may furnish an overall gloss of ‘information’, but we should not therefore be lulled into the assumption that all this ‘information’ is of a piece whatever its anchorage or provenance. The information/objects are constituted in and through their corres- ponding practices. It would be similarly misleading to assume that these diverse items might be aggregated in such a way as to be amenable to a formula in which ‘information’ can work as a single stable element – across constitutive practices –
not, at least, if we want to avoid a purely synoptic and remote ‘bird’s-eye view’ of the items informing the investing of trust.
Where EM departs further from rational choice, game and decision theorists is over the status and nature of the analytic technology to be deployed in describing trust. There is an argument as to whether these theorists’ characterization of trust as an outcome of explicit ratiocination, decision-making, or strategy could ever stand as a general account of how trust operates within the diverse weaves of ordinary activities and involvements. Nor in daily life do we invariably describe our lives as ‘a game’. Indeed, we ordinarily tend to use that noun as a reductive, relativizing term: ‘It’s only a game to him’, perhaps contrasting that with the serious commitments of others. As Edward Rose (1992) shows, such ordinary usages ineluctably shape the analytic ones: as analysts, we cannot just redefine ‘trust’ to fit our theoretical conveniences. I have elsewhere described this penchant to arbitrarily redefine ordinary-language terms as the ‘Humpty Dumpty tendency’ in sociology.
Even setting aside for a moment the status of the conceptual apparatus involved, to treat each and every investment of trust, or whatever, as describable in terms of Coleman’s model, or some other particular game-theoretic one, would be to commit one of the most familiar of philosophical fallacies, that of unwarranted extrapolation.
Nor is this fallacy simply a second-order one (though it is certainly that, too: see Anderson, 1987). Research I conducted some time ago with Wes Sharrock on simulation games indicated that issues of ‘reality’ and ‘realism’ are very much participants’ issues, too. In such games we see constant switches in the ‘accent of reality’ as expressed in conversational and other actions.2 The unwarranted extra- polation fallacy has a variety of aspects. For instance, as Alfred R. Louch (1966) points out, a game theory generalizes a particular motive-account that may well not be relevant outside game-playing, where winning or making a gain is not necessarily at issue. Furthermore, as Eric Livingston (2006) has argued in his studies of checkers, what makes a player interesting to play with has little to do with his or her motivation to win. There are many ways of winning, but only some are ‘fun’ for others, and thus sustain the game. Moreover, as Louch also points out, members may ascribe ‘rationality’ to an act even when it is not defined in terms of gain as an outcome, and David Houghton (1995) points out that prior calculative knowledge of some outcomes might devalue those very outcomes and thus devalues the ‘prior conditions’ from which a given outcome is derived.
The whole register of rationality, strategic game-playing, and so on, is problematic in the case of trust, especially when it works to present trust as a unitary, invariant phenomenon. The game frame of reference, when extracted from its specific contexts of ordinary relevance, loses its explanatory value and becomes impervious to disconfirmation. Game-analytic theories of trust ‘go down with the ship’, as it were, and it is up to us as critics to assist this process by providing the iceberg.
What kind of iceberg can we provide? In my view, there is no need to wait for what is ordinarily considered an empirical – let alone ‘scientific’ – determination of the matter. ‘In dealing with questions of this sort,’ according to Winch (1958: 18),3 ‘it is a matter of tracing the implications of the concept we use.’
A Conceptual Respecification of Trust
Following Winch, we strive for an elucidation of the concept of trust – an elucida- tion of its logical grammar in ordinary language – and for that we require what is essentially a philosophical ‘technique’ (of ordinary language philosophy – OLP) for tracking its implications.
Perhaps unexpectedly, it is Goffman who most succinctly sets up a first move in tracing such ‘implications’. This relationship between Goffman and OLP is currently a topic of interest in French philosophy and sociology. Perhaps in our pursuit of trust we can begin to locate it within what Goffman calls a ‘family’ of terms:
I think that at present, if sociological concepts are to be treated with affection, each must be traced back to where it best applies, followed from there to wherever it seems to lead, and pressed to disclose the rest of its family. Better, perhaps, different coats to fit the children well rather than a single splendid tent in which they all shiver.
(1961: xiii–xiv)
When applied to the concept of ‘trust’, Goffman’s advice is to capture its elusiveness or its shifting, fugitive qualities, rather than treating these as something of a problem. Moreover, it adverts us toward locating the concept of ‘trust’ within a set of ordinarily related terms that, in their relation to trust, allow us to explicate it in a non-reductive manner. We might suggest that ‘trust’, as an ordinary, laic concept, is situated in relation to other concepts, such as ‘undoubting’, ‘unques- tioning’. In this mode, we can, say, coherently refer to trust in someone as ‘placing faith in’ them, ‘placing them beyond doubt’, ‘having confidence in’ them or ‘believing in’ them. In all, this quotation from Goffman (written perhaps about the same time Garfinkel was writing his ‘trust’ paper) constitutes a welcome move toward the explicit focalizing of the logical grammar of concepts.
Rose (1992) gives us a diachronic analysis of ordinary uses of the English term ‘trust’ and notes that from the earliest natural uses it had expressed reliance, belief, faith and – earliest of all – fealty, fidelity or adherence to a person (cf. Seligman’s [2001] claims about the emergence of trust as a dimen- sion in the modern era). Only later did trust become more associated with con- fident expectation. In a sense, Rose’s approach affords us a working example of Goffman’s recommendation: to be sure, Rose himself (1992: 174) refers to ‘a close-knit extended family of words’ in relation to trust, many with meanings still
in ordinary use today. He gets closer than Goffman to explicitly seeing this family in terms of ordinary usage.
‘Trust’ can be situated in relation to other concepts (such as ‘faith’) and objects, too. J. Peter Rothe says about truck drivers:
A driver like Karl refused to be swayed by research studies that proved trucks with front-axle brakes more than halve stopping distances and they may help prevent jackknifing and spin-outs on ice and snow. Many truckers trust their experience more than they do science, clinging faithfully to their original beliefs.
(1991: 201)
Perhaps the next thing to note about variously conceiving of trust in terms of Goffman’s recommendation and Rothe’s observation is that confidence, faith, and so on, do not exclusively operate within the idioms of rationality, calculation or strategy. In this reading, trust may involve a ‘leap of faith’, even ‘going against one’s better judgment’, and so on.
Our approach, then, bears some elective affinities with the statement by Winch (1958: 44) that (linguistic) categories of meaning are logically dependent for their sense on the social interaction between individuals: on a community of laic conventional usage. Ordinary uses from within the family of concepts related to ‘trust’ may not fall unequivocally into any camp, either ‘rational’ or ‘non- rational’, but this, too, is provided for by Goffman and Rose.
The following is an excerpt – simply for illustration – from a BBC 2 TV interview between sports presenter Suzi Perry and a former British motorcycle racing champion, John Reynolds, after they witnessed a high-speed breakdown of rider James Toseland’s motorbike at a dangerous point at Brands Hatch racing circuit.
(RW: racingtrans:1)
1 SP: 2

7 JR:

But I mean it comes to the warm-up for example this morning you know, James is
trying different things to feel (0.1) more confident with the front (.) of that bike so
(0.2) do you feel now ‘sthough he::s gonna go out there and (.) have that (.) not
only the pressure of the home round but also now he’s got the problem of thinking
(0.2) oh my goodness is anythink gonna go wrong with my bike (0.2) or (0.1) can
you cut that out, is it possible to focus that much
Absolutely, that’s::s one thing you have to do (0.3) again yu know

(0.1) it’s one
9 10 11
  1. 13  SP:
  2. 14  JR:
of those things, the bike’s broken but er the team is a fantastic team (0.1) and they
supply fantastic machinery for the guys, and at the end o’theday you’ve just got to

trust what (.) is (.) around you and (.) just that (0.1) every time its gunna work and
(.) er (.) if you have the slightest negative attitude in world class racing (0.1) you

know, yuh may as well forget it
[ mm
[ so I’m sure James is well professional enough to just get on with it and (.) er (.)
put it (0.1) put the past behind him

We might
by rational choice theory – that is, rational grounds for trusting one’s racing motorcycle and the team that has prepared it (lines 8–10). But interspersed with that is the imperative (9–10) ‘you’ve just got to trust what is around you’ – something like ‘going beyond the information given’. Then there is the issue of ‘positive/negative attitude’ (line 11), which is hardly cast in the register of rationality, let alone formal rationality. In this respect, we can see that one facet of methodological irony involves the ironic formulation’s variance from members’ own cultural methods in their use of trust. The rational choice idiom, as a descrip- tion of the investment of trust, all too often sets up a competitive attitude with members’ practical reasoning on specific, local occasions, leaving the member looking deficient.

Garfinkel’s study of trust is one of his earlier works, and not everyone regards it as a fully fledged EM analysis. However, I have considered it well worth re-visiting in such detail. It is clearly an important part of the formation of EM.3 For me, it certainly remains by far the best study of trust amongst all the subsequent ones outside EM. Many of the more recent formal analytic studies of trust seem to do little more than survey each other.
Moreover, Garfinkel’s paper is one of the few studies of trust that departs from what Thomas P. Wilson (1970) glossed as the ‘normative paradigm’ (best represented by Talcott Parsons, 1963, 1975, in relation to trust) in sociology, in favor of a thoroughgoing treatment in terms of cultural knowledge-based sense-making practices.4 It also at least begins to replace the cognitivistic terms of reference to which so many other studies have subscribed – even though it is often interpreted cognitivistically by subsequent studies. For example, Giddens often describes the consequences of Garfinkel’s interventions in quite cognitivistic ways – at least, there is a cognitivistic residue in Giddens’ account. The language of psychology still pervades the research on trust, and Garfinkel’s study, for all its early status, gets away from this language and ushers in a more praxiological
discern in this excerpt the ‘practical version’ of some elements invoked
one. In this way it presages some of the more recent anti-cognitivistic arguments of EM, where with the rise of cognitive science these issues have gained far more prominence in recent years.
How, then, could we perhaps deal with the phenomenon of trust within the idiom of Garfinkel’s contemporary treatment of EM? Garfinkel’s work has always considered social phenomena as what we now call locally incarnate; that of course is what the original concepts ‘reflexivity’ and, perhaps more directly, ‘indexicality’ were all about. However, it can be argued that one way in which Garfinkel’s work has evolved is that it has come to incorporate a battery of concepts that further, maybe better, highlight this local character, concepts that of course were not in place in his original paper. Indeed, that this is so gives us the opportunity to deal with the multifarious situated manifestations of trust.
We might, as a prefatory orientation, look at two different natural mani- festations of trust (without according priority, as these illustrations are incom- mensurate). Firstly, we can examine explicit naturally occurring enunciated avowals and ascriptions of trust or trustworthiness – treating them as ‘wild phen- omena’ and asking: What are the local occasioning circumstances? What is the specific embedded determination of trust in each case? What are parties’ in situ methods for determining the specificity of sense of the term ‘trust’ in this case? How are such vehicles treated by recipients? What are the embedded, laic grounds for the investing or withdrawal of trust? How are these grounds adduced and managed endogenously, that is, within that selfsame setting? In all, what are the local ethno-methods (members’ methods) that are deployed in situ and in the living moment?
In these respects, a respecified version of some ‘takes’ on conversation analysis (CA) might well be of service, especially as it would help with the taking of the ‘linguistic turn’ in our approach to trust. In his study of trust, and in a subsequent joint paper with Sacks (Garfinkel and Sacks, 1970), Garfinkel writes of the centrality of the mastery of natural language to sense-making practices; perhaps this is a way of delivering that aspect of EM study. At the very least, con- versation-analytic transcripts, where possible, might support detailed noticings of the particular determinations of trust as phenomena of order. To be sure, the very discipline of writing a transcript leads the transcriber to scrutinize the data closely for minutiae of the phenomenon – an important way of addressing the issue of the phenomenal detail and phenomenal field properties of (in this case) trust. How, for instance, does an invocation of trust figure in a local cohort’s production of a situated moral profile for a given setting within that selfsame setting?
Trust Respecified: Some Perspicuous Instances
As members and participants in local orders of practice, we can note that expressions of trust/mistrust figure very specifically in such illocutionary avowals
and ascriptions, as part of particular local moral profiles: as, for example, in the following sequence from a legal tribunal investigating police corruption.
(Header: Tribunal on Police Corruption. A = Attorney; D = Defendant )
1 A:
3 D:

  1. 5  A:
  2. 6  D:
  3. 7  A:
  1. 10  D:
  2. 11  A:
  3. 12  D:
  1. 14  A:
  2. 15  D:
17 A:

19 D:

The position is that had there been money at Randolph Avenue and had
he kept his end of the bargain you would indeed have $25,000. No, I discussed with Sergeant Mckee and we felt it wasn’t the appropriate

thing to do and that you couldn’t trust Brien.
You mean you couldn’t trust him to keep his mouth shut?
No, that’s not what I’m saying.
You couldn’t trust him not to tell somebody that two members of the
Armed Robbery Squad had allowed him to keep $25,000 which might
very well be the proceeds of an armed robbery?
No, I’m not saying that.
What are you saying?
He was a person that would say one thing and then he would change his
mind and renege on what he was saying.
How does that bear upon your conduct?
Well, because of that I decided that it was inappropriate and I had to go
out and get that money.
Had he been a more reliable person, you would have let him keep the
money; is that what you are saying?
No, I’m not saying that at all. The circumstances of that .... (continues).6

This transcript of a tribunal on police corruption gives an illustration of the local work of attributing trust/mistrust in highly consequential circumstances. Coparti- cipants can be seen to be highly concerned not with a diffuse ascription of ‘trust- in-general’; instead, they show a concern for very specific, ‘targeted’ features of trustworthiness, or in this case untrustworthiness. The question for coparticipants is: How, and in what specific respects in this local instance, is the third party (Brien) not to be trusted? What, particularly, is the accused saying about the specific way in which the third party cannot be trusted; and in what specific respects in this local instance is he not accused of being untrustworthy? Note that this is a relational matter: the issue is not simply about Brien’s untrustworthiness but also about the accused’s dealings with him. Here, ‘trust’ is part of a language of social relations – part of a relational pairing, of membership categories, perhaps.
CA could greatly help us describe in instantiated ways the illocutionary vehicles in the ascription of trust which realize these differential rights.
Note, too, that, via the inspection of a naturally occurring case, we have located another member of the family of concepts concerning trust/ mistrust, namely ‘reliability’/‘unreliability’. Real circumstances can thus operate as ‘reminders’. We also have an issue of ascription of rights in the attribution of trustworthiness or otherwise. The defendant has first-person avowal rights, so that, for instance, the cross-examining attorney has to elicit an avowal from the defendant, or can propose the particular way in which Brien is untrustworthy. However, it is the defendant who can, for instance, accept or deny a particular proposition. It is the defendant who can ascribe a particular formulation of Brien’s untrustworthiness and ‘make it stick’. (Of course, there are institutional constraints on the process, but in this case they seem manifestly to be extensions of ordinary practice, e.g., a first person-warranted ascription.)
We might propose that considerations of logical grammar and families of concepts concerning trust add something to Garfinkel’s early argument, some- thing that is compatible with – though still, regrettably, implicit in – his later respecifications of EM. We might, indeed, also suggest that one interpretation of Goffman’s recommendation that concepts be ‘followed through to wherever they seem to lead’ is that the logical grammar of the concept ‘trust’ be exhaustively mapped out with regard to specific local instances (pace Winch).
The other main focus of trust is how it is built as a tacit phenomenon into the routine production of a given local order. How is a given determination of trust incorporated into that production as an integral feature? Asking questions in this way might restore a concern for the particular case, rather than the attitude of disdain for it that Ludwig Wittgenstein (1953) decried in philosophy. It will move us further away from formal analytic concerns with trust-in-general, trust abstractly conceived, formally described and without reference to lay members’ reasoning. After all, note in the sequence above how co-participants themselves show an orientation to the ‘fine-tuning’ of the concept of trust to a particular complex of phenomenal detail.
One way of looking at trust is possibly through focalizing ‘found’ per- spicuous settings, where a particular setting may turn out to have a propaedeutic quality for the researcher: that is, the very phenomenal cast of this setting renders it particularly accessible and ‘grasp-able’ to the researcher. With regard to trust, this approach would at least allow us to avoid a priori definitions, as is indicated in this statement by Sacks, cited by Garfinkel (2002: 182). Wishing to avoid aprior- ism in making a particular distinction that he senses (in fact, ‘possessables’ versus ‘possessitives’, that is, abandoned vs stolen cars as they ‘appear’ to the police in everyday police practice), Sacks declared:
... but I don’t want to write definitions. Instead, I want to find a work group somewhere, perhaps in Los Angeles, who, as their day’s work, and
because they know it’s their day’s work, will be able to teach me what I could be talking about as they know it as the day’s work.
Sacks found a local, natural setting that turned out to be particularly and relevantly instructional, (the setting concerned police handling of abandoned vehicles) in which a suitable membership pair was used in the course of their day’s work. We may say that Sacks wished to find a laic determination of the kind of membership category pair he had in mind. Of course, there is a potential issue concerning the formal generalizing of such a situated use.
Thus, we may find ordinary settings that, in relative terms, render generic ‘trust’ available for EM study. A first inspection, based on some of my own researches, suggests at least the following ‘possibles’ – traffic flow, particularly the locally organized serial ordering of high-speed traffic on motorways, or alternatively a series of US Forest Service vehicles traveling in close order along mountain or forest tracks in emergency conditions. Other locally organized serial orders, such as pedestrian flow-files in urban public spaces, and queues or waiting lines, furnish further opportunities for the study of trust. In these cases the so-called ‘information’ oriented to by participants is setting-specific (and practice-specific), relating to the distinguishing particularities of the evolving gestalt contextures in terms of which these local orders are constituted. The term ‘information’ is often held to imply an interchangeability of such details; this is most certainly not what is meant here. Since the details are not interchangeable, the formal analytic rubric is not appropriate for the explication of these local orders.
Such orders are participant-produced orders, realized through reciprocally sensible, locally embedded practice. As we shall see in the examples below, and as Anne Rawls (2005) has argued, temporal ordering is a constitutive feature of such produced settings: the sequential ordering of the setting is one major way in which such ordering is worked out as an identifiable course of action. This temporal ‘working out’ of a setting involves trust in ‘underwriting’ the reciprocities involved amongst the production cohort (A.W. Rawls, 2005: 180). Some local studies of trust have already been initiated, notably those of Gonzalez-Martinez (2001 and n.d.), who, instead of conceiving of trust as an ‘across-the-board’ or ordinary phenomenon, refers to ‘practical trust’ which is, essentially, tied to local production and local-natural accountability. The site upon which she examines the operation of trust as a praxiological process is that of a pre-trial court hearing to decide whether or not a trial is, in fact, required.
Now for some abbreviated examples. Trust in local orders of traffic flow consists in presupposed (but monitored) reciprocities in the very production of that particular serial (and thus temporal) order, particularly in the competent and attentive maintenance of a locally established standard pace and standard interval with no sudden changes through braking: the smaller the interval, the more ‘evident’ the trust involved. Also, standard trajectories are reciprocally ‘presupposed-but- monitored expectancies’ to which drivers are held through various sanctions.
These are lay conceptual matters. A particularly perspicuous example is, perhaps, files of racing cars or racing motorbikes traveling along straightaways within inches of each other at around 200 mph, and sometimes cornering in file at speeds of 160 mph or more (with conventions, too, for safely changing places in that serial order). Clearly, too, what Sacks calls ‘members’ measurement systems’ are elements in the exhibiting and maintenance of trust, particularly as trust as a local phenomenon is always trust pro tem – the warrantability of trust as monitored over the temporal course – and may be revised at any point in that course, and may work to recast that warrantability either retrospectively or prospectively, or both.
Similarly, serial orders of pedestrians moving through public spaces place trust in locally established standard paces and trajectories, as well as in co- participants’ identities; indeed, the issues of pace, trajectory (exhibited, temporally organized course) and identity turn out to be closely related and are, together, constitutive of the flow-file or the queue. Garfinkel (1963: 238) defines a trust- worthy person as someone who can produce a public show of respect for pre- scribed attitudes, and visibly exhibiting the prescribed metrics for identifiably assembling such local public orders exemplifies that.
In the ‘tradition’ of Garfinkel’s interventions, we might as a purely supple- mentary approach look at breaches of trust, though in the spirit of the quotation from Sacks given above, we might seek out naturally occurring, naturally situated breaches rather than contrived ones. Even then, the study of breaches of trust should indeed at best be only a supplementary one, for as Rose puts it:
Doubt interrupts and is rare. Trust commonly prevails. ... Trust as a rule is the rule, doubt the exception. As a rule people have to do with things through the exercise of trust pared by doubt occasionally, hardly ever by the exercise of doubt relieved occasionally by trust.
(1992: 170)
(This quote also suggests the danger of tying trust invariably with risk.)
With this caveat in mind, we might make some suggestions about the study of breaches of trust. We can begin by asking how an attitude of distrust, suspicion or doubt come to be locally and naturally occasioned; we might perhaps start up such a study by respecifying Sacks’ (1972) analysis of police observations and assessment of moral character: that is, how trusted identities of persons – identities naïvely presented and received – come to be re-documented by the

police ‘on the spot’ as suspicious, possibly criminal.
Sacks refers to the identification at a certain point in time within local

circumstances of a trait or deed that appears to the police observer as incon- gruous with the respectable or trusted identity. This, in turn, occasions a redocu- mentation on the part of the police of the ‘actual identity’ of the observed person. Sacks progressively specifies an ‘incongruity procedure’ whereby this identity redocumentation, this withdrawal of a trusted identity, is brought about.
This procedure, he observes, is a refinement of a lay person’s methods. Thus, for instance, where police are looking for possible betting activities of an illegal nature:
... if the individual gives them a ‘double look’, they’ll check him. By this, the officers mean that if an individual sees them in their unmarked car and then turns to look at them once again, chances are that the individual has some gambling paraphernalia on his person ... .
(Sacks, 1972: 287; see also Paperman, 2003: 405–9)
Distrust or suspicion, then, involves a transformative operation, where at some point in time during an observation: ‘... police transform information concerning the paths of activity the observed person selects into a description of a set of acts which may be seen as transacting an offence’ (Sacks, 1972: 290). I feel that this approach is readily amenable to respecification in contemporary EM terminology.
We might also refer to the occasioned categorial order of trust and mistrust, relating to what Lena Jayussi (1984) has termed ‘disjunctive category- sets’ for a given person (‘businessman’ at t1 and ‘drug dealer’ at t2) to do with ‘categorizational asymmetry and disjunction’ (Jayussi, 1984: 122–31, though Jayussi tends to flatten the temporal ordering involved). Perhaps, too, we might categorially locate a diffusely present ‘attitude of potential mistrust’ within the membership category ‘police officer’, where incumbents may be on the alert for specific circumstances that realize such an attitude. In so doing, we can continue to locate trust in the context of laic sociological description and sense-making – an issue that Sacks addressed from the earliest phases of his work. Categorizations – or ‘membership categorizations’ in particular – were, for Sacks, especially clear examples of such laic sociological description, and the re-categorization of a person from ‘businessman’ to ‘drug dealer’ is a prime instance of re-description – one that, plausibly, involves a withdrawal or reduction of trust. The methods (Sacks, 1972) members employ for, say, co-selecting membership categorizations, choosing between categorizations, substituting categorizations, and so on, emphasize the taking of the ‘procedural turn’ (Sharrock and Watson, 1988), which is integral to the EM treatment of all phenomena of order – including, of course, trust. We are moving toward a firmly proceduralized description of trust.
Concluding Comments
In this paper, I have tried to offer a reminder of what I feel to be one of the most formative early studies in ethnomethodology, coincidentally addressing a sub- stantive phenomenon that has come to gain massive attention in contemporary social and political science. To me, despite the amount of subsequent work in social and political science, Garfinkel’s study remains far and away the best, and
for this reason alone richly deserves not to be overlooked. On the other hand, the ‘logical grammar’ consideration of Garfinkel’s early analysis indicates that a respecification of his conception of trust may also be appropriate, and that he may have over-stabilized this conception. In my view, his 1963 study stands in need of a ‘grammatical’ approach as articulated by the later Wittgenstein and those in his tradition, and, in a somewhat differing way, in the work of Gilbert Ryle. However, it may be the case that Garfinkel’s later reconceptualizations of his position (Garfinkel, 2002) have worked to more effectively highlight and extend his lifelong commitment to a conception of social order as purely locally avail- able. It may consequently be the case that his later reconceptualizations (though not inspired by Wittgenstein) are more amenable to ‘grammatical’ and ‘family resemblance’ considerations.
That trust has gained such significance affords us an anchorage for specific comparisons of the methodological asymmetries between EM studies and formal- analytic research technologies. By that selfsame token I hope to have indicated the incoherence that can only come from the assimilation of EM within formal analytic studies of trust or, worse, the attempt to contain EM studies within formal analytic ones or the use of EM as a stylistic patina for what is essentially a formal analytic study. In all the formal analytic discussions of Garfinkel’s article on trust, I have not read one that shows sufficient appreciation of the fact that it represents a methodologically radically different and incompatible program, or properly follows through the implications of this. Nor do they appreciate the impli- cations of the argument for constitutive orders, objects and information – and their relationship to trust and other constitutive requirements. This and related observations have, in turn, allowed us some pointers toward an EM-based critique of formal analytic concerns. Finally, I have tried to suggest some analytic resources for an updated EM treatment of trust.
I have received much valued help and advice over two drafts of this paper from friends and colleagues, not least the participants in the International Institute for Ethnomethodology and Conversation Analysis Conference ‘International Perspectives’ at Bentley University, August 6–9, 2005, where a much earlier version of this paper was presented. In particular, I wish to mention: Tim Berard, Andrew Carlin, Jeff Coulter, Jon J. Dreissen, Andrew J. Goldsmith, Christian Greiffenhagen, Sheena Murdoch, Ged Murtagh, Louis Quéré, Rupert Read, Wes Sharrock, Yves Winkin and Maria Wowk. Some of these do not, or will not, fully agree with the argument or the form or order of its presentation. I can only proffer the traditional claim that all defects are my own responsibility. A much earlier version of this analysis was published in French in 2006 (Watson, 2006). Finally, special thanks to Anne Warfield Rawls for scrupulous reading and central suggestions for revision of this paper.
  1. It is the strangest misapprehension of Garfinkel’s study of trust that it somehow cannot deal with conflict. Misztal (1996) is one author who sustains this misapprehension.
  2. For some empirical indications of this, see Sharrock and Watson (1985, 1987), and Watson and Sharrock (1990). On analytic issues, see also Anderson (1987) and Sharrock and Watson (1985).
3. I thank Tim Berard for prompting me to bring these Winchean considerations into plainer view.
  1. Contemporary CA, too, even when writing of preference organization, seldom if ever refers (even contrastively) to Garfinkel’s study. I believe that a discussion of possible elective affinities (or not) is well overdue and would be most profitable.
  2. This alternative program is dubbed by Wilson (1970) ‘the interpretive paradigm’. This is a fine designation except that, so far as ethnomethodology is concerned, we are not dealing with ‘interpretation’, nor is it a ‘paradigm’. (Indeed, Karl Mannheim’s original notion of ‘the documentary method of interpretation’ suffers from the same drawback.) Pace the claims of symbolic interactionists, we do not go about constantly ‘interpreting’ our world. ‘Interpretation’ only occurs in very specific circumstances; by and large, we apperceive our everyday world as ready-made, not needing ‘interpretation’. And, as I have noted elsewhere in this paper, the logical grammar of the term ‘interpretation’ involves a downgrading in a scheme of credibility and doubt, or even a relativization.
  3. I thank Andrew J. Goldsmith for this trial transcript. I have re-transcribed it as best I can given its original format, for ease of reading. Despite this transcript being on the public record, I have chosen to adhere to the confidentiality guidelines prevalent in social sciences and have changed the names of coparticipants and places. The attorney is cross-examining a police officer (Defendant) about possible financial misdeeds on his part vis-à-vis a criminal under his observation.
Anderson, R.J. (1987) ‘The Reality Problem in Games and Simulations’, pp. 43–50 in D. Crookall, C.S. Greenblat, A. Coote, J. Klabbers and D.R. Watson (eds) Simulation-Gaming in the Late 1980s. Oxford: Pergamon Press.
Coleman, J.S. (1990) Foundations of Social Theory. Cambridge, MA: The Belknap Press of Harvard University Press.
Durkheim, É. (1933) The Division of Labor in Society, trans. George Simpson. Chicago: Free Press. (Orig. pub. 1893.)
Gambetta, D. (1988) ‘Can We Trust Trust?’ pp. 213–37 in D. Gambetta (ed.) Trust: Making and Breaking Co-operative Relations. Oxford: Basil Blackwell.
Garfinkel, H. (1963) ‘A Conception of, and Experiments with, “Trust” as a Condition for Stable Concerted Actions’, pp. 187–238 in O.J. Harvey (ed.) Motivation and Social Interaction. New York: Ronald Press.
Garfinkel, H. (1967) Studies in Ethnomethodology. Englewood Cliffs, NJ: Prentice- Hall.
Garfinkel, H. (2002) Ethnomethodology’s Program: Working Out Durkheim’s Aphorism (edited and introduced by Anne W. Rawls). Lanham, MD: Rowman and Littlefield.
Garfinkel, Harold (2008) Toward a Sociological Theory of Information (edited and introduced by Anne W. Rawls). Boulder, CO: Paradigm. (Orig. distributed 1952.)
Garfinkel, H. and H. Sacks (1970) ‘On Formal Structures of Practical Action’, in J.C. McKinney and E.A. Tiryakian (eds) Theoretical Sociology: Perspectives and Developments. New York: Appleton-Century-Crofts.
Giddens, A. (1990). The Consequences of Modernity. Cambrdige: Polity.
Giddens, A. (1991).
Modernity and Self-Identity. Cambridge: Polity.
Goffman, E. (1961).
Asylums. New York: Doubleday Anchor. Gonzalez-Martinez, E. (2001) ‘Comment agir en confiance avec un partenaire
dont on se méfie?’, Réseaux 108, Special Issue ‘La Confiance’: 87–124. Gonzalez-Martinez, E. (n.d.) ‘The “Practical Trust” Produced during Judicial
Hearings’. Unpublished ms, Sociology Department, Boston University. Heritage, J. (1978) ‘Aspects of the Flexibilities of Natural Language Use: A Reply
to Phillips’, Sociology 12: 79–103.
Houghton, D. (1995) ‘Reasonable Doubts about Rational Choice’,
70: 53–67.
Jayussi, L. (1984)
Categorization and the Moral Order. London: Routledge &
Kegan Paul.
Livingston, E. (2006) ‘Ethnomethodological Studies of Mediated Interaction and

Mundane Expertise’, Sociological Review 54: 405–25.
Louch, A.R. (1966)
Explanation and Human Action. Berkeley: University of
California Press.
Luhmann, N. (1979)
Trust and Power. Chichester: Wiley.
Misztal, B.A. (1996)
Trust in Modern Societies. Oxford: Blackwell.
Paperman, P. (2003) ‘Surveillance Underground: The Uniform as an Interaction

Device’, Ethnography 4(3): 397–419.
Parsons, T. (1963) ‘On the Conception of Political Power’,
Public Opinion
Quarterly 27: 116–51.
Parsons, T. (1975) ‘Social Structure and the Symbolic Media of Interaction’, in

P.M. Blau (ed.) Approaches to the Study of Social Action. London: Open
University Books.
Rawls, A.W. (2005) ‘Garfinkel’s Conception of Time’,
Time and Society 14(2/3):
Rawls, J. (1955) ‘Two Concepts of Rules’,
The Philosophical Review 64(1):
Rose, E. (1992)
The Werald. Boulder, CO: The Waiting Room Press.
Rothe, J.P. (1991).
The Trucker’s World. New Brunswick, NJ: Transaction. Sacks, A. (1972) ‘Notes on Police Assessment of Moral Character’, pp. 280–93 in
D. Sudnow (ed.) Studies in Social Interaction. New York: Free Press. Schutz, A. (1967) The Phenomenology of the Social World, trans. G. Walsh and
D. Lehnert. Evanston, IL: Northwestern University Press.
Schwartz, H. and J. Jacobs (1979)
Qualitative Sociology: A Method to the Madness.
New York: Free Press.
Seligman, A.B. (2001) ‘Complexité du rôle, risque et emergence de la confiance’,

Réseaux 108, Special Issue on ‘Trust’ (‘La Confiance’): 37–62.
Sharrock, W.W. and D.R. Watson (1985) ‘Reality Construction in L2 Simulations’, pp. 42–57 in D. Crookall (ed.) System 13(3) (Special Issue: ‘L2 Simulation Applications in L2 Education and Research’): 42–57.
Sharrock, W.W. and D.R. Watson (1987). ‘“‘Power” and “Realism” in Simulation and Gaming’, pp. 35–43 in D. Crookall, C.S. Greenblat, A. Coote, J. Klabbers and D.R. Watson (eds) Simulation-Gaming in the Late 1980s. Oxford: Pergamon.
Sharrock, W.W. and D.R. Watson (1988) ‘Autonomy among Sociological Theories: The Incarnation of Social Structures’, pp. 56–77 in N.G. Fielding (ed.) Actions and Structure. London: Sage.
Simmel, G. (1950) The Sociology of George Simmel, trans. and ed. by K.H. Wolff. New York: The Free Press.
Simmel, G. (1978) The Philosophy of Money, trans. T. Bottomore and D. Frisby. London: Routledge.
Sorokin, P.A. (1956) Fads and Foibles in Modern Sociology and Related Sciences. Chicago: Regnery.
Watson, R. (2006) ‘La Confiance Comme Phenomène Pour La Sociologie’, pp. 143–68 in A. Ogien and L. Quéré (eds) Les Moments de la Confiance. Paris: Economica.
Watson, D.R. and W.W. Sharrock (1990) ‘Realities in Simulation/Gaming’, pp. 231–8 in D. Crookall and R. Oxford (eds) Simulation, Gaming and Language Learning. New York: Newbury House.
Wilson, T.P. (1970) ‘Normative and Interpretive Paradigms in Sociology’, pp. 80–103 in J.D. Douglas (ed.) Understanding Everyday Life: Towards the Reconstruction of Sociological Knowledge. Chicago: Aldine.
Winch, P. (1958) The Idea of a Social Science and Its Relation to Philosophy. London: Routledge & Kegan Paul.
Wittgenstein, L. (1953) Philosophical Investigations, trans. G.E.M. Anscombe. Oxford: Basil Blackwell.
Rod Watson is a member of LIAS (Institut Marcel Mauss, Paris). He has written extensively on subjects related to ethnomethodology since the early 1970s. For the past decade he has spent considerable time lecturing and doing research in Paris, and he has made an important contribution to the French discussion of these issues.
Address: LIAS-IMM, 54 buld. Raspail, 75006 Paris, France. [email:]