Revisiting the Relationship between Arguing and Convincing: Towards a New Pragmatic Account

Abstract

How do individuals change their minds as a result of argumentation? It is generally assumed the speech act of argumentation can trigger a change of mind in the other party—the perlocutionary act of convincing. This means that a discussant changes her commitment relative to the proposition under scrutiny when the other party presents argumentation that is in some way convincing or persuasive. I challenge this received view by showing that argumentation cannot trigger this change of commitment in the way that scholars commonly assume. Convincing cannot be triggered by assertives that are already in the listener’s commitment set, nor can it be triggered by assertives that are newly introduced in the discussion. Using the notion of “joint commitment” I propose an alternative account according to which change of mind is the result of two speakers jointly experiencing facts as stipulated by a joint commitment. I conclude the paper by sketching the impact of such an approach in the study of argumentation and provide suggestions for further developments.

1. Two Problems

The Fourth International Colloquium in the Philosophy of Science was held in July 1965 at Bedford College, London. Anybody who was anybody attended, including Karl Popper, Thomas Kuhn, Imre Lakatos, Alan Musgrave, Stephen Toulmin, Paul Feyerabend, and many others. Owing to the fundamental differences between the views held by these philosophers, the resulting discussions must have been thrilling. It was in such an atmosphere that Thomas Kuhn exclaimed:

How am I to persuade Sir Karl, who knows everything I know about scientific development and who has somewhere or other said it, that what he calls a duck can be seen as a rabbit? How am I to show him what it would be like to wear my spectacles when he has already learned to look at everything I can point to through his own.

(Kuhn 1970, p. 3)

Kuhn’s remarks were not related to any specific puzzle in the philosophy of science. Rather, they constituted a more general expression of the difficulties involved in setting off a gestalt switch within the confines of a relatively short exchange during a colloquium. In argumentation-theoretical terms, what Kuhn is deploring is the predicament that himself and Popper share too much common ground. How can Kuhn produce any change of mind in someone who practically knows everything he (Kuhn) knows? Convincing under conditions of too much common ground seems difficult because there is hardly any asymmetry from which the change of mind can originate.

However, convincing might seem difficult to achieve also from an opposite scenario. For if there is too much common ground, there can also be too little. This is highlighted in the following passage from Kekes’ discussion of fundamental differences between social groups:

For what can a right-to-life advocate say that would persuade a militant feminist, a gay liberationist to a moral majoritarian, a champion of law and order to a lawyer specializing in getting criminals acquitted on technicalities, a Mormon to a hippie, a marine to a transcendental meditator, or, for that matter, a philosophy professor to a junkie? The moral sensibilities of these people are so far apart that there is no common ground for one even to explain to the other his or her position.

(Kekes 1993, p. 7)

Here we have the opposite problem: the parties involved share too little common ground in order for one party to have a genuine chance of convincing the other. In fact, as Kekes points out, the parties are so far apart it hardly ever comes down to individuals engaging in an argumentative discussion. Epistemologists and argumentation theorists sometimes refer to this as deep disagreement (Popa 2022; Lavorerio 2021; Ranalli 2021). We thus have two problems related to the notion of convincing: there can be too much common ground (situation described by Kuhn) or too little (situation described by Kekes). In both cases, the changing of the mind implied in the term convincing, the change in dialectical commitments, seems to be impeded.

However, do these problems touch upon the notion of convincing itself, conceptually speaking, or are they just extreme cases? Do these situations say something about how minds are changed, or can we simply shrug our shoulders and conclude that convincing is just not going to happen in those specific situations? Intuitively, one is perhaps inclined to go for the latter explanation. After all, we do not typically experience problems with the phenomenon of convincing in our day-to-day interactions. Convincing someone of a point of view, or being convinced ourselves, seems to come naturally. If it does not happen, then the arguments were not up to par, or the process was somehow disturbed, but convincing itself, as a perlocutionary act, is possible and it is connected to the act of argumentation. When people disagree, they can argue their way to agreement. No pragmatic conundrums are involved.

In this paper, I want to take a closer look at this traditional pragmatic connection between the speech act of argumentation and the perlocutionary effect of convincing. I will not deny the obviously correct observation that people actually change their minds (commitments) relative to propositions as a result of interacting with others. This phenomenon is pervasive and unquestionable. What I want to challenge is that this change is an effect of the act of argumentation, that it follows somehow from the act of being “convincing” or “persuasive”. The predicaments experienced by Kuhn and Kekes are real and relevant, and they suggest that we must reconsider our traditional account of how arguments change minds. I will propose an alternative pragmatic reading of the situation in which speakers change their minds as a result of joint experiences. In this account, the act of argumentation serves to discover the disagreement (rather than resolve it) and resolution comes through joint experiences of mutually recognized facts.

2. Convincing as an Effect of Argumentation?

The claim that arguments can convince forms the cornerstone of much work in argumentation theory, informal logic, rhetoric, and other disciplines concerned with rational communication between individuals (Johnson and Blair 1994; van Eemeren et al. 2014; Walton 2008). The act of convincing is characteristically seen as the effect of argumentation, the two standing in an illocution–perlocution relationship to one another (van Eemeren and Grootendorst 1992; van Eemeren and Snoeck Henkemans 2016). Of course, not all arguments are equally suited to achieve the effect of convincing. However, some are. Additionally, a central challenge in the pragmatic study of argumentation has been to understand the conditions under which an argument possesses this quality of being convincing.

How does this illocution–perlocution relationship between argumentation and convincing actually work? Assuming that argumentation is characteristically achieved through the speech-act category of assertives, how does performing assertives lead to changes in the other party’s commitment? The general answer seems to be this: an argument is convincing, that is, rationally persuasive, when the hearer’s change of commitment is triggered by the propositional content of the assertives employed and not by other pragmatic features that these assertives might have (see e.g., Blair 2012; Dutilh Novaes 2021; Johnson 2000; Lynch 2012; van Eemeren and Grootendorst 2004; Woods and Walton 1982). This means that, in order for convincing to occur, the change of mind must be based on what the arguer is saying not on how she is saying it. In putting forward argumentation, I might sway you with my baritone voice or my charming body language such that you change your mind, but then no convincing has occurred in the strictest sense.

It will be useful from this point onwards to work with a case in point. In her recent study of rational persuasion, Dutilh Novaes (2021) asked the basic question that I am tackling here, namely, “What does it take for a Skeptic to be persuaded by a proof?”. After surveying the many different answers that have been given to this question, Novaes concludes that convincing (or rational persuasion, as she calls it) must be understood in a dialogical framework as a discussion between a Prover and a Skeptic. Her answer to the question is as follows:

-

To become convinced of the conclusion of a proof, the Skeptic must accept its premises.

-

He must not be in possession of counterexamples, either global or local.

-

He must also deem each step in the proof/argument to be individually perspicuous and convincing (Dutilh Novaes 2021, p. 52).

Duthil Novaes later explains this game between the Prover and Skeptic in more detail:

[…] assuming Skeptic does grant the premises proposed by Prover and no global counterexample is found, then Prover proceeds to put forward a sequence of further statements that she claims follow necessarily from what Skeptic has already granted. If Prover is right that these statements follow necessarily from what has been granted, then Skeptic is indeed committed to them […] and thus may not refuse to grant them. If eventually the intended conclusion is reached through successive inferential steps that were not questioned or refuted by Skeptic, or if Skeptic’s objections and requests for clarification have been dealt with satisfactorily, then Prover will have succeeded in her goal of persuading Skeptic of the conclusion.

(Dutilh Novaes 2021, p. 53)

The problem is that the Kuhn-problem appears in this definition of convincing. For if the Skeptic does indeed accept all the premises employed by the Prover as condition 1 stipulates, how can there still be disagreement between the two and how can the Prover be said to have changed anything in the Skeptic’s commitment set? Contrariwise, if the Skeptic does not accept the premises as condition 1 stipulates, then by the same definition there is certainly no convincing involved—in which case we arrive at the Kekes-problem.

Let us give a general formulation of this two-pronged situation in order to pinpoint more concretely how the problems arise. We start with two parties disagreeing on a proposition.

Step 1. Prover and Skeptic disagree on p.

The disagreement between Prover and Skeptic can differ in scope and intensity. What matters here is simply that Prover and Skeptic do not have the same commitments relative to p. The Skeptic, we can imagine, would say “No” when asked whether she agrees that p, while the Prover would say “Yes”. Next, the Prover needs to make sure that the statements she is about to employ in her argumentation are granted by the Skeptic (otherwise the disagreement increases instead of being resolved). We know that, in real-life situations, parties tend to skip this step and jump to the argumentation part directly—implying as it were that the premises employed in the argumentation should be acceptable to the other party. Yet in order to make all components of the convincing process visible, this explicit route will be necessary. Thus, with $\phi$ as the collection of statements making up the Prover’s premises, we have the next step:

Step 2. Prover asks Skeptic to grant a set of statements $\phi$.

The formula $\phi$ stands for what we normally call “the argumentation” (or “the defense”) advanced by the Prover. It is at this stage not necessary to say more about how the statements in $\phi$ are interconnected. The structure within $\phi$ can be as simple as that of the age-old syllogism and as complex as any argumentation structure we come across in everyday interactions (Snoeck Henkemans 2000). In fact, the reader is free to substitute $\phi$ with any definition or representation of an argument in any formal or non-formal language available (deduction, induction, abduction, defeasible logics, etc.). The argumentation does not have to be of a certain kind in order to count as argumentation, or, as Hamblin put it, “The actual logical relation between premises and conclusion of an argument may be anything at all” (Hamblin 1970, p. 230). However, it is important that $\phi$ includes both the propositions (e.g., q, r, s) and what we would see as inference rules from those propositions to p (e.g., q & r & s $\to$ p).

This is then the crucial moment, for it all comes down to the Skeptic’s position relative to $\phi$. Following Dutilh Novaes’ account, in order for the Skeptic to be convinced, the following would have to occur:

Step 3. Skeptic grants $\phi$, is in no possession of counterexamples and finds all steps conspicuous and convincing.

Sure enough, there is a version of this condition in just about any account of argumentation that is based, more or less loosely, on some logical or dialectical model of convincing. However, are we to conclude that the Prover “will have succeeded in her goal of persuading Skeptic of the conclusion” (Dutilh Novaes 2021, p. 53)? As agreed, persuading involves a change of commitment, a transformation of conflict into consensus. Yet this is not the situation we are looking at. For the Skeptic already grants $\phi$ without the Prover advancing anything in defense for $\phi$ or indeed performing any other additional speech act relative to $\phi$. It seems more natural to say that the Skeptic was already convinced of p or, put differently, that the Prover and Skeptic simply discovered that there was not any disagreement between them after all. In natural language, their conversation could have developed as follows:

Prover: You should close the door (p)

Skeptic: Why should I?

Prover: Do you agree that it is your turn to close the door and that in all cases where it’s your turn, you should close it? ($\phi$)

Skeptic: I agree with all that ($\phi$)

*Prover: So, there you have it! (p)*

*Skeptic: You’re right (p)*

In such a situation we would not normally describe the Prover as having convinced the Skeptic, but rather as having reminded, or in some other sense made the Skeptic (more) aware of their already-existing consensus. Notice that the Prover does not do more than offer $\phi$ for assent and $\phi$ is immediately accepted (as the Dutilh Novaes account requires). It is therefore the information asymmetry, not the disagreement, that the Prover eventually solves by her act. The Skeptic does not have less knowledge or different commitments than the Prover, she just does not remember her commitments as much (or as fast) as the Prover does. Now, there is, of course, change involved in reminding someone of something: the Skeptic is at t₁ in a state of mind that is different from her state of mind at t₀ in that, roughly speaking, the Skeptic was aware of her commitment to p at t₁ but not at t₀. Yet this is not the kind of commitment we want to associate with the perlocutionary act of convincing. After all, when one is convinced, it is not the awareness regarding p that needs to change, but the acceptability of p. The other possibility at Step 3 is, of course, that the Skeptic does not agree with $\phi$. In this case, it is uncontroversial that no convincing has taken place.

To sum up, we can formulate the problem presented here as a dilemma. After the Prover presents the Skeptic with $\phi$, then we have to ask ourselves what the status of $\phi$ is in the Skeptic’s commitment. Two scenarios are possible, neither of which incorporates the traditional conception of convincing:

Scenario 1

The Skeptic already agrees with $\phi$, in which case there was no disagreement involved after all because both $\phi$ and p are accepted by the Skeptic. Through assertives, the parties discovered that they were in fact in agreement. No convincing took place because there was too much agreement on starting points.

Scenario 2

The Skeptic already disagrees with $\phi$. By advancing $\phi$, the Prover increases the disagreement from p to (p & $\phi$). No convincing takes place because there is not enough agreement on starting points.

The reader will have noticed that Scenario 1 corresponds to the Kuhn-problem (“How can A convince B of something when each knows everything else the other does?”) and Scenario 2 corresponds to the Kekes-problem (“What can A say to B when the two are fundamentally different in their starting points?”). In some models of argumentation such as the pragma-dialectical one, scholars have insisted that the parties should be allowed to introduce new starting points halfway through the discussion (van Eemeren and Grootendorst 1982, 2004). However, we see that this amounts to an exchange of one horn of the dilemma for the other. For now, regarding the newly introduced starting points, we have to ask: does the other discussant agree with these starting points or does she not? If she does, then there was not any disagreement after all, and we are in Scenario 1; if she does not the disagreement is now increased, and we are in Scenario 2. There seems to be a mismatch between the easily recognizable social practice of convincing and the many models of convincing that are somehow resulting from the combination of logic (conclusion being derived from the premises) and dialectic (premises being granted by the other party).

3. Changing Minds and Joint Commitments

When our models tell us that some practice we seem to observe everywhere is in fact impossible, there are two ways out. Either we give up our interpretation of that practice as falling within the scope of that model, or we give up our models. I started this paper with the assumption that people do change their minds (i.e., their commitments) as a result of argumentation, so for the present paper, the first option is not available. Regarding the second option, I want to propose a model based on the notion of joint commitment (Gilbert 2000, 2014). Joint commitments as defined by Gilbert have many interesting features but the most basic one relevant here is that a joint commitment is a form of answerability that involves two or more people and that cannot be created or rescinded unilaterally (Gilbert 2014, pp. 40–42). The situations we typically associate with individuals being convinced by arguments are, I will argue, better understood in light of joint commitments.

In situations falling under Scenario 1 above, there is already a joint commitment between the Prover and the Skeptic, and the Prover reminds the Skeptic of this joint commitment.

1	Prover: You should close the door (p)
2	Skeptic: Why should I? (φ)
3	Prover: Do you agree that it is your turn to close the door and that in all cases where it’s your turn, you should close it? (φ)
4	Skeptic: True, I forgot that we agreed to that (φ)

Of course, forgetting a joint commitment is one of the many explanations for the Skeptic’s reply in line 2. The Skeptic might, e.g., want to check whether the Prover is herself remembering their agreement, or which joint commitment she selects in this situation. There are plenty of reasons to have your discussion partners clarify the joint commitment to which they are appealing. Of course, it can take a while before the two actually find the joint commitment that is needed coordinate actions. For example:

1	Prover: You should close the door (p)
2	Skeptic: Why should I? (φ)
3	Prover: Do you agree that it is your turn to close the door and that in all cases where it’s your turn, you should close it? (φ)
4	Skeptic: Is it my turn? (φ)
5	Prover: Well, it’s Wednesday! (β)
6	Skeptic: Is it Wednesday? (β)
	etc.

These are all situations falling under Scenario 1 where there is no convincing because there is no actual disagreement (Kuhn-problem). The exchange of arguments serves to discover the joint commitment that the two need to coordinate their activities—whether it is closing the door or something else.

Let us now turn to Scenario 2. The notion of joint commitment can also be employed to explain how minds can change in Scenario 2 where there is no sufficient agreement on the acceptability of propositions (Kekes-problem). Here as well we have to move away from the idea that argumentation itself triggers the change of mind. Nevertheless, the argumentation can lead to the establishment of a new joint commitment on the basis of which one of the parties can change their mind regarding p. Crucially, however, this new joint commitment cannot pertain to the acceptability of a proposition or set of propositions $\phi$—for otherwise we would land right back into Scenario 1 in which the parties discover that the apparent disagreement was just a misunderstanding regarding a forgotten joint agreement. Instead, this new joint commitment must pertain to an outside test, authority, or judge to which both parties agree to delegate the resolving of the disagreement. What I have in mind is a situation such as the following where the Prover and the Skeptic formulate a joint commitment to undertake a “fact-checking” procedure:

1	Prover: You should close the door (p)
2	Skeptic: Why should I? (φ)
3	Prover: Do you agree that it is your turn to close the door and that in all cases where it’s your turn, you should close it? (φ)
4	Skeptic: Is it my turn? (φ)
5	Prover: Well, it’s Wednesday! (β)
6	Skeptic: Is it Wednesday?
7	Prover: We can check the calendar!
8	Skeptic: Let’s do it!
9	[Prover and Skeptic check the calendar and it is indeed Wednesday]
10	Skeptic: I’ll go close the door!

In the past, scholars have sometimes stipulated that such situations fall outside the scope of convincing—that in such situations the dispute is settled rather than resolved. For example:

The resolution of a difference of opinion is not the same as the settlement of a dispute. A dispute is settled when, by mutual consent, the difference of opinion has in one way or another been ended—for example, by taking a vote or by the intervention of an outside party who acts as a judge or arbitrator. Of course, reaching a settlement does not mean that the difference of opinion has really been resolved. A difference of opinion is only resolved if a joint conclusion is reached on the acceptability of the standpoints at issue on the basis of a regulated and unimpaired exchange of arguments and criticism.

(van Eemeren and Grootendorst 2004, p. 58)

According to this view, then, creating a joint commitment (“by mutual consent”) regarding an outside intervention in the dispute is not a form of resolving the disagreement. In other words, if we agree to the joint commitment of consulting the calendar or the rabbi then we are abandoning our rational pursuit since we quite leave the dispute to these external parties, the calendar or the rabbi. However, given that in Scenario 2 the Prover and the Skeptic do not share enough starting points to do anything else, then the stipulated distinction is rather unfortunate. Without any past joint commitments to reveal that the apparent disagreement was just a misunderstanding, the only way for the Prover and the Skeptic to do something about their disagreement is to try to tackle it through a new joint commitment regarding an outside intervention (see also Popa 2022). In fact, in the example where they disagree on what day of the week it is, the only rational approach is to create a new joint commitment that the dispute is to be decided by looking at the calendar. If two discussants argue back and forth about what day of the week it is, we hardly praise them for keeping to the process of an “unimpaired exchange of arguments and criticism”. This is then how a new joint commitment “Let us check the calendar on this” can lead to the parties undertaking the action mentioned in the commitment and the result of that action can lead to one of the parties having changed their mind. Of course, the two might discover that the joint commitment is insufficiently clear and that they need to specify their joint action further. However, once such issues for extracting the evidence have been decided, and assuming the evidence is not in some sense inconclusive, one of the two will have changed their mind. The Skeptic will have been convinced, not by the Prover, but by the “facts” that were observed according to their joint commitment.

To sum up, if the parties find themselves in Scenario 1 (Kuhn-problem), then argumentation can lead to the discovery of a (forgotten or overlooked) joint commitment in which case the parties revert to an earlier form of agreement. No disagreement, hence, no convincing. If the parties find themselves in Scenario 2 (Kekes-problem), then argumentation can lead to the discovery that no such joint commitment exists in which case the parties can create a new joint agreement pertaining to an outside intervention that can decide the matter. The disagreement was real, and, if the evidence was conclusive, one of the parties will be convinced. However, in neither of the two cases is the act of argumentation itself the one triggering the effect of changing a discussant’s mind.

4. Discussion

The account presented here raises several important questions regarding the relationship between argumentation and convincing. While it is not the purpose of the present paper to work out all the methodological consequences of the change of perspective proposed, some of the more urgent issues deserve immediate consideration.

First, we should take seriously the possibility that the Kuhn-problem and the Kekes-problem are not the only two available options for convincing, as I suggested in the beginning, but rather extreme points on a continuum of options between sharing too many and too few starting points. A typical in-between case would be one in which the Prover asks the Skeptic to grant some (not all) of the starting points needed to reason from one granted premise to the conclusion. Then the Prover adds some new starting points and in this way the convincing occurs on the middle ground between no shared starting point and too many shared starting points. One could perhaps insist that the Prover is still reminding the Skeptic of her commitment to p since the “new” starting points are indeed accepted by the Skeptic right away, but as the logical distance between p and $\phi$ grows, this becomes increasingly implausible. Let me explain.

In the simplest case, or nearly so, $\phi$ simply consists of one proposition (q) and the claim that this proposition implies the conclusion (q → p). Together, these two propositions (“q” and “q → p”) make up $\phi$. However, with $\phi$ being so structurally simple, we are clearly facing the Kuhn-problem now: if the Skeptic does indeed accept $\phi$ consisting of those two statements, how can she ever be said to disagree on the acceptability of p? Yet as the defense $\phi$ becomes more complex, that is, when there are many propositions and inference rules between q and p, then there is presumably a connection that the Skeptic does not “see” or “know” right away, and the argumentation can be said to perform this step. In this case, indeed, it would be something of a stretch to insist that this still falls under the heading of remembering. Plato famously advanced this surprising explanation in Meno where Socrates does not teach the slave how to double the area of a square, but he (Socrates) merely reminds him of it. The calculations needed to solve the problem are presumably carried out by working with nothing, but premises are known to the slave. This explanation is possible in a Platonic theory of knowledge because of the notion of an immortal soul which acquires knowledge before we were born and is only reminded of this knowledge during our lifetime. Yet I take it that we have to explain the relationship between argumentation and convincing within a simpler ontology. The point, therefore, remains valid: when there is a great distance between q and p (e.g., q → r → s → t → w → p), it becomes increasingly implausible (as it is in Meno) to say that the Skeptic is simply being reminded of her commitments. For example:

1	Prover: President Jones will not run for a second term (p)
2	Skeptic: Why is that?
3	Prover: Because there have been so many shootings lately (q) Skeptic: I don’t see the connection …
4	Prover: Well, the president is a conservative (t)
5	Skeptic: And?
6	Prover: Conservatives support gun rights (s)
7	Skeptic: So?
8	Prover: Well, these gun rights are why you have so many shootings (r)
9	Skeptic: So?
10	Prover: Can’t you see? The president is in favor of a policy that is nowadays responsible for so many shootings. He would never risk face-loss by running for a second term knowing that people would oppose him just for his gun activism and have him lose etc. (w) Skeptics: I see your point and I agree Jones will not run for a second term (p)

In such examples, the complexity of the relationship between q and p makes it implausible to maintain that the Prover is simply reminding the Skeptic of her commitment to $\phi$, i.e., her commitment to “q → r → s → t … → p”. There is, at least intuitively speaking, more going on. Nevertheless, given that the Skeptic accepts all the Prover’s assertions without the Prover arguing for them, there is still a sense in which the Skeptic accepted those assertions before the Prover presented them (otherwise, if she had not, then the Prover’s asserting should not do much to alter their status). Perhaps then remembering is not the right pragmatic description of what is going on, but nevertheless, the Skeptic was in a real sense already on board with p because she was already committed to $\phi$. On the continuum between the Kuhn-problem and the Kekes-problem, this is in any case closer to the former.

Second, it could be argued that the account of convincing advanced here restricts the concept’s application to the admittedly limited cases in which the Kekes-problem occurs (i.e., there is actual disagreement) and it can be solved by a joint commitment to facts. This restriction does indeed occur, but it is less harmless than it might seem. For two parties the term “fact” can in practice cover a wide range of external interventions: the calendar, the rabbi, the latest statistics, the Internet, the dictionary, the university professor, the math textbook, etc. Furthermore, some disputes might not need an external authority of this sort to decide the matter. If the Prover and the Skeptic disagree on whether the Prover can do twenty push-ups, then there is little left to do but to decide the matter by doing the twenty push-ups. Indeed, there is hardly any restriction on what two individuals can consider to be deciding “facts of the matter” given the wide variation in epistemic cultures (Knorr-Cetina 1999) that we see throughout institutions and cultures. This is not to say, of course, that there is no rationality in perception and that experience cannot be hijacked (Siegel 2017).

What happens when the Kekes-problem is encountered but there are no such facts? Does the present account fail in the case of deep disagreements (Popa 2022; Lavorerio 2021; Ranalli 2021)? As I have argued elsewhere, deep disagreements do not form a separate category of disagreements but rather occur whenever the Kekes-problem occurs (Popa 2022). It follows that our most complex and divisive social issues regarding constitutional rights, social problems, political issues, and moral dilemmas do not constitute a phenomenon qualitatively different from the versions of the Kekes-problem discussed here (e.g., where the Prover and the Skeptic cannot agree on the day of the week). The main difference is this: whereas the Prover and the Skeptic in the examples given here could rely on a calendar to decide the issue, and quite naturally do so through a new joint commitment, there are no corresponding agreed-upon procedures for individuals disagreeing on questions of power, liberty, justice, and other values. Convincing is unlikely to occur on those issues and thus the concept does not really apply to those situations, because if the Kekes-problem is indeed discovered through well-directed acts of argumentation as in our examples, it is unlikely that the parties could formulate a joint commitment for an experimentum crucis of the kind given in the example with the calendar.

5. Conclusions: Towards a New Pragmatic Account of Argumentation

Let me sum up the points made above before I flesh out some of the consequences of this perspective. I started with the observation that the illocutionary act of argumentation is saddled with a perlocutionary act it prima facie cannot perform—the perlocutionary act of convincing. Why is this particular illocution–perlocution relationship problematic? Argumentation that is based on already-accepted starting points cannot convince because if the starting points are indeed already accepted, there is no disagreement to begin with. I referred to this as the “Kuhn-problem”. By contrast, argumentation that is based on starting points that are not accepted cannot convince because there is no acceptability to transfer from the starting points to the conclusion. I referred to this as the “Kekes-problem”. Because of these two problems, I concluded that there must be something different underneath that social practice we can surely take for granted, namely that individuals do change their minds as a result of argumentative interactions with others.

As an alternative account, I proposed that what discussants are doing in argumentative interactions where a change of mind occurs is one of two things. Either they arrive at a joint commitment they already have, thereby discovering that the apparent disagreement was no disagreement after all; or they arrive at a dead-end from which they can be saved by a new joint commitment regarding a decisive test (unless, of course, they decide to abandon the whole thing and move on despite their disagreement). In both cases, the exchange of arguments serves to discover, define (and refine) the parties’ positions relative to one another. The positions are in this sense “interactionally emergent” (Jacobs, Jackson, and Zhang, this volume). The parties do not start with a fully formed difference of opinion, they start with a bit of apparent misalignment, and they seek, through an exchange of arguments, to find the origin of this misalignment. Convincing, then, in the sense of a radical switch from A to non-A in one’s commitment set, only occurs in the case of a Kekes-problem solved by a joint commitment relative to external facts. This is when the parties do something together that decides the matter.

While this conclusion goes against standard argumentation-theoretical work in which resolving and setting a dispute are distinguished as two pragmatically different phenomena (van Eemeren and Grootendorst 2004), I do not think the idea itself is particularly new. In fact, this train of thought can be easily traced back to Plato’s Euthyphro, where it is noted that at least some discussions are such that they can only be settled by a joint experience of facts rather than through discourse:

SOCRATES: Let us look at it this way. If you and I were to differ about numbers as to which is the greater, would this difference make us enemies and angry with each other, or would we proceed to count and soon resolve our difference about this?

EUTHYPHRO: We would certainly do so.

SOCRATES: Again, if we differed about the larger and the smaller, we would turn to measurement and soon cease to differ.

EUTHYPHRO: That is so.

SOCRATES: And about the heavier and the lighter, we would resort to weighing and be reconciled.

EUTHYPHRO: Of course.

SOCRATES: What subject of difference would make us angry and hostile to each other if we were unable to come to a decision? Perhaps you do not have an answer ready but examine as I tell you whether these subjects are the just and the unjust, the beautiful and the ugly, the good and the bad. Are these not the subjects of difference about which, when we are unable to come to a satisfactory decision, you and I and other men become hostile to each other whenever we do?

EUTHYPHRO: That is the difference, Socrates, about those subjects. (Plato, Euthyphro, 7b–c).

In the same way, Perelman and Olbrechts-Tyteca (2013) delineated the realm of rhetoric as pertaining to those discussions that could not be settled through an appeal to empirical observations or the demonstrative sciences such as mathematic or formal logic. Of course, by the time Perelman and Olbrechts-Tyteca were writing, this observation served to broadening the by-then restricted realm of rhetoric so as to include many fields not normally assigned to it, fields such as science, law, politics, business, etc. However, perhaps without realizing it, they were continuing a tradition that supports the distinction between discussions settled by discourse and those settled non-discursively, by the joint experiences of seeing, doing, hearing, etc. Once the distinction was in place, however, argumentation theorists largely focused on the former category—understandably so, given that argumentation theorists are trained, broadly speaking, as analysts of discourse. However, the result of this focus had the surprising effect of restricting the field of argumentation to cases where the dispute cannot be decided (“settled”) through an experimentum crucis. Why is this surprising? If the Prover and the Skeptic disagree on, e.g., how many chairs there are in the next room, then if they decide to walk to the next room and decide the dispute through the joint experience of counting chairs, the event will be of little or no interest among argumentation theorists, rhetoricians, informal logicians, and the like. However, if they keep at it, through a “regulated and unimpaired exchange of arguments and criticism” (see Section 2), then the interaction is suddenly argumentation-theoretically relevant. However, surely the two are variants of the same phenomenon, namely attempting to do something about our disagreements.

Although I have not attempted to provide a full-fledged pragmatic notion of argumentation here based on this reunification of discursive and non-discursive approaches to disagreement, I think that the Kuhn–Kekes dilemma forces us to widen our perspective on argumentative interaction to include situations where it is ultimately the experience of facts, of what the parties agree to regard as “facts”, that triggers changes of mind. Yet I am afraid that even this is an old idea in philosophy. This is essentially the conclusion at which Augustine arrives in his dialogue The Teacher, namely that one “learns not from words but from the things themselves and his senses” (12.39, 15). The perlocutionary act of convincing thus can only be consistently related to the experience of the “things themselves”, those which the parties agree, through joint commitment, to see as decisive facts. The argumentative interaction preceding this experience is only the preparatory work.