ISSA Proceedings 1998 – The Normative Structure Of Adjudicative Dialogue
If you ask them, most people will say that disagreements should be resolved through dialogue. If you ask them what this means, however, you are less likely to get a straightforward answer. While commitment to dialogue as a mode of conflict resolution is widespread, most of us are less than clear about what this commitment entails. What does it mean, exactly, to discourse dialogically?
In the heat of discursive contestation, we tend to focus on the matter at issue, and attend little, if at all, to the normative structure of dialogue itself. This contributes, I think, to a general lack of clarity concerning the norms in question. Here theory can aid practice by shedding light on the norms that govern adjudicative discourse. By stepping back from particular disputations and articulating the otherwise tacit knowledge that underlies and structures them, the theorist can sharpen and reinforce basic intuitions about the process. In this paper, I aim to show that resolution-oriented discourse has a distinctive normative structure that is partially subject to theoretical explication.
It is not an ethic of disputation, but a logic of disputation, that I am after here. I am interested in how various dialectical gambits alter the structure of obligations and alternatives that disputants face as the dialogue unfolds. Like any other logic, a logic of disputation must strike a balance: it must capture some of the richness of the practice being modeled, yet still cast core structures into bold relief; it must be relevant to concrete discursive contexts, yet abstract away from the particularities of such contexts; it must do justice to the complexity of reason-giving discourse, while bringing simplicity and clarity to our understanding of it.
A good way to reconcile these constraints is to model reason-giving discourse as a kind of game. After identifying a useful typology of moves, we clarify the conditions under which moves of each type are permitted. Finally, we characterize the normative implications of each move-type in terms of its “effects” on the distribution of discursive commitments and entitlements. Such a logic, I believe, can facilitate what Robert Brandom calls “deontic scorekeeping” – the keeping track of discursive commitments and entitlements (Brandom, 1994). Since this is an important part of resolution-oriented discourse, a logic of disputation can actually enhance our capacity to resolve disagreements dialogically.
Dialogical disputation begins when one party to a discussion expresses disagreement over, or an inability or unwillingness to go along with, some claim or assumption made by another. Before describing the process that ensues, we need to identify the dialectical resources available to the disputants. Already we know something important about this, for in order even to disagree, the interlocutors must first share a language. Donald Davidson has shown us that, to share a language, people must share a large number of beliefs in common (Davidson, 1984). To share a language is also to jointly recognize a large number of what Brandom calls “material inference proprieties” (Brandom, 1994). Put simply, there will be a variety of inferential transitions that both participants will be predisposed to recognize as appropriate or valid. We can depict this shared background – or “common ground,” as I like to call it – using a Venn diagram, as the intersection of two necessarily overlapping belief-sets. It is to elements of this set that participants must ultimately appeal in their attempts to gain dialectical leverage.
There are three kinds of inference relevant to our story: permissive, committive, and incompatibility.[i] Two sentences are related by permissive inference whenever entitlement to one entitles one also to the other. For example, entitlement to ‘There is smoke rising from yonder chimney’ generally carries entitlement to ‘There is a fire in the fireplace beneath’ (barring some unusual circumstance like the presence of a firetruck outside the building). Two sentences are related by committive inference whenever commitment to one commits one also to the other. For instance, commitment to ‘Fido is a dog’ commits one also to ‘Fido is a mammal.’ Two sentences are said to be incompatible whenever commitment to one precludes entitlement to the other. For example, commitment to ‘The sea is green’ precludes entitlement to ‘The sea is colorless.’
The centrality of the notions of commitment and entitlement in this account makes it a good idea to define them more precisely. To be committed to a claim is to be obliged to defend it discursively (in the standard case, by presenting supporting reasons or evidence), if appropriately challenged. One undertakes discursive commitments primarily by asserting, and one divests oneself of discursive commitments by withdrawing or renouncing claims formerly asserted. Roughly speaking, then, one is committed to whatever one has asserted and not withdrawn, plus whatever follows from these, via committive inference. (It is worth mentioning that, in real life, actions other than speech-acts can also carry discursive commitments. Showing up late can mean that one has some explaining to do, for example. While the non-discursive undertaking of discursive commitments is an important phenomenon, and well worth further study, it will not occupy me further here.)
To be entitled to a judgment is basically to be [rationally] permitted to employ it discursively: to use it as a premise, for example. (Or to act upon it in other appropriate ways – here again we encounter an important bridge between the discursive and the non-discursive.) Some entitlements are presumptive: players begin the game with default entitlements or judgments that do not stand in need of justifying. These judgments are properly thought of as rationally in order unless some reason for doubting them surfaces. Other judgments must be redeemed or justified if entitlement to them is to be had. The stock of entitlements, then, begins the game non-empty, and expands whenever judgments are redeemed or justified, and shrinks whenever judgments are undermined or refuted.
It is not easy to identify the precise contours of the stock of default entitlements. Fortunately, it is not necessary to pin this down. For interlocutors invariably bring to the encounter their own unique conception of what one is entitled to assume, assert, and take for granted. If they are open-minded and persuadable, the dialogic encounter will often compel them to alter this conception: to add judgments that appropriate reason-giving performances can secure, and to delete judgments that appropriate reason-giving performances can undermine. In this way, interlocutors are compelled to take on new commitments and abandon old ones: to in effect modify the substance of their conceptions of entitlement.
Because the norms that govern rational discourse reinforce commitments that can be dialogically upheld and put pressure on commitments that cannot, they are well-suited to bringing about convergence in belief. It is their nature to foster consensus regarding what one is entitled to and what one ought to be committed to. Jurgen Habermas is right when he writes that the ideals of mutual understanding and unrestrained consensus are embedded in the norms that govern reason-giving discourse (Habermas, 1984: 287ff). To say that a certain kind of discourse has resolution or consensus as its “inherent telos,” however, is not to say that it serves these ends at the expense of truth. It is certainly possible for the interest in agreement to conflict with the interest in truth, but the two interests can also coincide. If we assume that both parties to our dialogue are honest and sincere, open-minded and persuadable, and that both want the dialogic encounter to render a verdict that is as close to the truth as they can make it, then we can treat the strictly non-identical dialogical ends of resolution and truth as functionally coincident.
In the skeletal version of the reason-giving language-game to be articulated here, there are five basic move-types: assertion, challenge, defending, withdrawal, and concession. An assertion puts a claim forward and commits the person making the assertion to it. A challenge is directed against an assertion or assumption (called its “target claim”), indicating disagreement or irresolution, and suspending entitlement to its target, pending an investigation. A challenge puts the claimant in the position of having to defend the claim or assumption challenged. A defending is meant to redeem an entitlement that has been suspended or is at issue. A withdrawal renounces a previous commitment, taking the claim withdrawn “off the table.” A concession essentially withdraws a challenge, signaling the challenger’s newfound willingness to accept the point at issue.
It proves analytically useful to subdivide the categories of challenge and defending. Some challenges present one or more reasons against the claim they target. For example, a prosecutor might confront a defendant with eyewitness accounts that contradict his testimony. Following Lance (1988: 59ff), I call such challenges “assertional.” The reasons against I call grounds for doubt. The idea here is to present assertions that are (a) difficult to deny, and (b) incompatible with the target claim, in order to undermine entitlement to the target claim. In general, assertional challenges take the form ‘How can you say that P, when R?’ where P is the target and R the grounds for doubt. (It simplifies matters to focus on the case where R is the lone ground for doubt.)
The other possibility is that a challenge presents no such grounds for doubt. Again, following Lance, I call these challenges “bare.” Bare challenges simply ask the claimant to provide reasons for the claim targeted. A bare challenge might take the form ‘How do you know that P?’ where P is the target-claim. (Note that this typology of challenges is analytically complete, for a challenge must either present grounds for doubt or not, there is no third option.)
We must also distinguish two kinds of defending. A defending is a move intended to redeem entitlement to a claim at issue, and typically follows on the heals of a challenge. What I call a “direct” defending takes the challenge it is a response to as well-posed, and presents reasons that purport to redeem entitlement to the claim it targets. The idea is to present assertions that (a) permissively entail the target-claim, and (b) are themselves entitlements. An “indirect” defending, by contrast, attempts to show that the challenge it is a response to is somehow misposed (for example, that it presupposes falsely) – that it ought not to be regarded as suspending entitlement. In effect, an indirect defending is a challenging of a challenge. (Note here that this typology, too, is analytically complete, for a defending must either accept the challenge as well-posed or not, there is no third option.)
Dialogical disputation commences when one party (“the challenger”) issues a challenge and the other party (“the claimant”) responds with a defending rather than an immediate withdrawal. By attending closely to the type of challenge and the type of defending, we can characterize the structure of the normative “field” – the pattern of obligations and options that will structure the ensuing dialogue. There are four possibile combinations of opening moves: a bare challenge followed by a direct defending, a bare challenge followed by an indirect defending, an assertional challenge followed by an indirect defending, and an assertional challenge followed by a direct defending. I will examine them in that order.
Suppose, then, that a challenger issues a bare challenge and the claimant elects to defend directly. For example, ‘How do you know that a second trimester fetus is a living human being?’ might prompt the direct defending: ‘Well, a fetus has a heartbeat and brain activity by ten weeks.’ Here the claimant has presented what supporting reason for the claim at issue. The question we must ask is this: what effect does such a “move” have on the structure of the normative field?
The primary effect is to change what is immediately at issue. The question of the target claim’s epistemic status is temporarily set aside, and attention shifts to the supporting reasons. Two questions arise about them: First, is the claimant entitled to these claims? Second, do they confer entitlement on the target claim? These questions correspond, respectively, to what traditional argumentation theory would call the explicit and suppressed premises of the claimant’s justificatory argument. These now take center stage, and the issue becomes whether they are tenable. If the challenger concedes that both are tenable – that is, if he is unwilling to challenge the claimant on either point – then he is rationally obliged to concede the target claim. If willing to challenge the claimant on either point, however, he needs to do so. In terms of our example, he might reply, ‘It is not clear to me that a heartbeat and brain activity are enough to make a fetus a human being.’ Here the challenger has targeted the inference from the supporting grounds to the target claim, indicating that there is a question about whether that inference is in fact entitlement-conferring. The suppressed premise that articulates the inference (specifically: ‘If a fetus has a heartbeat and brain activity, then it is a human being’) becomes the new point at issue, and the original question awaits the outcome of the embedded issue.
As challengers and claimants avail themselves of certain dialectical options, and neglect others – offering this supporting reason rather than that, or targeting this assumption rather than that one with a challenge – they together navigate the discussion out onto one or another branch of a vast “game-tree,” which can be thought of as representing all the different ways the dialogue might play out. As the interlocutors steer the discourse out onto the branches on the tree, unresolved issues are bracketed, and embedded issues are taken up. This corresponds to movement “up” the tree – away from the “trunk” and towards the branches. Resolutions are effected when this movement is reversed: premises and/or inferences on which embedded issues hinge prove mutually agreeable, and logic compels either a concession or a withdrawal. Open branches of the game-tree are in this way closed off, embedded issues are resolved, and disputants can return to the embedding issues (represented here as larger supporting branches) with an expanded common ground – a broader basis for building consensus.
Let us see turn now to the case of a bare challenge followed by an indirect defending. Suppose, in other words, that a challenger issues a bare demand for evidence and the claimant wishes to contend that such a demand is inappropriate or misposed. Can a bare challenge, which hazards no grounds for doubt, be misposed?
Certainly. Bare challenges represent a powerful move in the language-game because they purport to saddle the claimant with a justificatory burden without exposing the challenger to any comparable risk. (Because no grounds for doubt are involved, the challenger does not need to worry about defending them.) Yet this power comes at a price: the right to issue bare challenges must be theoretically circumscribed or the game becomes unbalanced. The regress skeptic, who seeks to defeat all claims to knowledge by simply iterating bare challenges until supporting reasons give out, is the hypothetical embodiment of the fact that the reason-giving language game breaks down if the right to issue bare challenges is not kept within appropriate bounds.
Bare challenges are only appropriate when the burden of proof is on the claimant. For if the claim challenged is presumptive, or reasonably taken for granted, it is not up to the claimant to provide reasons for the target claim, it is up to the challenger to provide reasons against it. It takes an assertional challenge, in other words, to undermine a presumption. Hence bare challenges that target presumptions are misposed. For example, the bare challenge ‘How do you know that you have ancestors?’ is inappropriate for falsely presupposing that the onus is on the claimant. ‘How do you know that the earth has existed for many years past?’ is similarly misposed.
Indirect defendings to such bare challenges can take a very simple form: ‘It seems probable enough. Why do you doubt it?’ Such a response neatly shifts the onus back where it belongs (or seems to the claimant to belong). Typically, such a move will elicit a brief indication of why the challenger understands the target claim to be unworthy of being asserted. In effect, this amounts to his sharing his grounds for doubt – to the challenger’s replacing his bare challenge with an assertional one. The dialogue can then unfold as outlined below. The interesting case here occurs when the challenger does not take on the onus of disproof, but simply insists that the burden lies with the claimant.
To prevent our dialogue from degenerating into a series of refusals to shoulder the onus of proof (or disproof), we need to stipulate a non-subjective measure of presumptiveness, or immunity to bare challenge. My proposal is this: if the claim at issue is reasonably likely, given the information available to both interlocutors (i.e. as reckoned on the common ground), then the claim is presumptive, and the onus is on the challenger. If not, the claim is non-presumptive, and the onus is on the claimant.
The virtue of this proposal is that it instructs disputants that might otherwise remain at loggerheads to begin sharing the elements of their background knowledge that incline them to think that the claim at issue ought, or ought not, to be treated as presumptive. Typically, this will uncover the bit of background knowledge possession of which leads one to think of the claim as prima facie reasonable (or unreasonable), and the non-possession of which leads the other to think the opposite. When this happens, the question of onus has already been resolved: the onus did lie on the possessor, but the sharing of it shifted the onus to the disputant who just learned of it. If it turns out that it is the claimant who possesses the decisive bit of information, then her sharing of it should be regarded as a direct defending of the claim at issue, and the dialogue should proceed as outlined above. If it turns out that it is the challenger who possesses the decisive bit of information, then his sharing of it should be regarded as an assertional challenge, and the dialogue should proceed as outlined below.
So let us turn now to disputations that begin with assertional challenges. The distinctive feature of an assertional challenge, of course, is its grounds for doubt. An example would be: ‘Your claim that our moral dispositions are largely inherited seems unlikely, given their extreme malleability.’ The primary normative effect of such a challenge is to change what is immediately at issue. Once again, the question of the status of the claim it targets gets put on the back burner, and attention shifts to the grounds for doubt. Two questions about them arise: First, is the challenger entitled to these grounds for doubt? Second, are these grounds for doubt in fact incompatible with the target claim? As in the case of direct defendings, these questions correspond to the explicit and suppressed premises of an argument – only this time it is a falsifying argument based on an incompatibility, rather than a justifying argument based on an entitlement-conferring inference.
If the claimant feels compelled to concede both the grounds for doubt and their incompatibility with the target – that is, if she is unable or unwilling to challenge either one or the other – then she is rationally obliged to withdraw the target claim. In terms of our example: ‘Perhaps moral dispositions are not inherited after all.’ On the other hand, if she is willing to challenge the challenger on either point, she needs to do so. The result is an indirect defending of the target claim. She might do so in this case by saying: ‘But the malleability of our moral dispositions is itself adaptive, and provides further evidence of their genetic basis.’ Here the claimant has raised the question of whether ‘Our moral dispositions are malleable’ is genuinely incompatible with ‘Our moral dispositions are largely inherited.’ This, then, becomes the new point at issue – a new branch of the game tree will have opened up – and until a concession or withdrawal closes it off, the issue will remain unresolved.
When a claimant offers a direct defending in response to an assertional challenge, she attempts to redeem the target claim by providing reasons for that simply outweigh the assertional challenge’s reasons against. For example, if my claim that Bill was in Copenhagen last month is challenged with: ‘But Bill said he was not going to Copenhagen,’ I could respond directly by saying: ‘Yes, when I talked to him there, he told me that his plans had changed.’ Here the stronger evidence of Bill’s presence in Copenhagen simply overwhelms the weaker evidence of his absence. Direct defendings in such cases are functionally equivalent to direct defendings offered in response to bare challenges, and the norms that govern their appraisal differ in no serious particular.
This completes my survey of the four possible combinations of opening moves and their normative implications. The branches of the game-tree continue to multiply and expand outwards from here, and though I cannot discuss all of them, the branching process and the norms at work are precisely those I have already discussed: the basic structure of challenge and response is the same, though the point at issue will have shifted. If disputants can keep tack of where they are in the game-tree (which branch they are on), remember which entitlements are at issue, and which entitlements hinge on which others, then when their shared background begins to compel withdrawals and concessions, branches of the game-tree will close, and simple logic will compel the adjustments to commitments and conceptions of entitlement that will move the disputants toward consensus.
This is not to say that consensus or agreement is guaranteed, for the common ground may be insufficiently extensive to draw forth the requisite concessions and withdrawals. Pride, ego, or closed-mindedness may also prevent resolution. But if there is enough common ground, if the disputants approach the exchange in the right spirit, and if they are clear about the normative structure of adjudicative dialogue, they stand a good chance of resolving their dispute amicably, and broadening their understanding of the matter at issue.
i. Here again I follow Brandom.
Brandom, Robert (1994). Making It Explicit. Harvard: Harvard University Press.
Davidson, Donald (1984). Inquiries Into Truth and Interpretation. Oxford: Oxford University Press.
Habermas, Jurgen (1984). The Theory of Communicative Action, vol. 1. Boston: Beacon Press.
Lance, Mark (1988), Normative Inferential Vocabulary: The Explicitation of Social Linguistic Practice. Diss. University of Pittsburgh.