1 Introduction

Any deep disagreement is a serious difficulty for argumentation. The problem is not that a deep disagreement cannot produce a justified belief or action at all—it is that it cannot be dealt with by the truth-seeking processes of argumentation. Fogelin’s (1985) classic characterization of these disagreements is that they are “disagreements, sometimes on important issues, which by their nature, are not subject to rational resolution” (p. 7). The key features of such disagreements, as summarized by Godden (2013), are that they lack:

two contextual features that characterize normal (or near-normal) arguments:

(i) they lack background context of broadly shared beliefs and preferences and

(ii) they lack existing, shared procedures for their resolution. (p. 1)

“Procedures” here refers primarily to the reasoning procedures of introducing evidence, weighing it, and making logical inferences, rather than to dispute resolution procedures at the level of choosing a jury trial, mediation, negotiation, or working within other social practices and institutions for handling disagreement.

The concern is what to do if a disagreement turns out to be “deep”. Must we give up on it no matter how much we might need an answer? Argumentation theory hopes to demonstrate that argument is the best route to resolving disagreement. If we cannot use argument, either the disagreements must go entirely unresolved, or they must be handled by other means such as persuasion, coercion, or fiat, which have less authority to convince us.

The core of the debate about deep disagreements has been whether there are any disagreements which are undecidable even for the best-informed, most logical reasoners. For example, Lugg (1986) contends that there is no significant limit to reason. Disputes are not inaccessible to reason just because disputants lack shared background context or existing, shared procedures for their resolution. They need not begin with shared background because they may be able to create it through reasoning. Phillips (2008) similarly argues that it is possible to reason across significant differences in foundational beliefs and in social position related to age, gender, ethnic background and similar attributes. Adams (2005) proposes that it is important to continue reasoning even in a deep disagreement: we may be in situations which oblige us not to give up, for example, medical treatment decisions where there is a time limit on what is possible, or situations where we have the moral and legal obligation to reach a decision as a group.

There are other factors besides context and procedure which may inhibit or prevent good reasoning, but the hope has been that these additional factors can be dealt with outside the argument itself. Examples of external factors include affective and social conditions. For instance, an affective issue, such as fear of the other participants, can turn a moderate disagreement into an unbridgeable gulf because some people refuse to participate. There may also be social circumstances in which argument cannot even start because it is unacceptable to question, let alone reason with, authorities.

Procedural difficulties have been treated as internal to the argument, but in fact overlap into external factors. For instance, a procedural clash could either be about how to decide which evidence is relevant, or whether to include less capable reasoners in the discussion. The first is internal, the second is external, but they overlap if the issue is whether some kinds of evidence can come only from less capable reasoners.

In effect, the discussion of deep disagreements has compartmentalized core beliefs and reasoning processes as factors internal to a deep disagreement, separate from any external factors. Pragma-dialectics offers one example of this compartmentalization. Describing the conditions for productive discussion, Van Eemeren and Grootendorst (1988) separate the conditions into three “orders” or levels:

The discussion rules are then called first order conditions, the internal characteristics which specify a reasonable discussion attitude are second order conditions, and the external requirements of the circumstances in which the discussion takes place are third order conditions. (p. 287, italics in original)

The “external requirements of the circumstances” come from the social setting in which the discussion takes place:

…In order for persons to be able to reason, and to bring to their reasoning an attitude of willingness to express their opinions and listen to others, the psycho-social reality in which the individuals operate should be such that it fulfils the third order condition that everyone has the right to advance his view to the best of his ability… [and] that he has a real voice on the subject and is not, for example, totally dependent on the compassion of the person whom he is addressing. (p. 287)

One justification for this compartmentalization is that if we want to be able to focus on whether a disagreement can be resolved using reason, we do not want the reasoning process to be hampered by any fallibility of the actual arguers. Feldman and Warfield (2010) make this explicit: to explore differences of opinion between equally well-informed and competent experts we must acknowledge that experts may not in fact be precisely equally well-informed or equally competent, but for purposes of understanding the reasoning we can presume that they are peers, where:

In the stipulative sense…peers literally share all evidence and are equal with respect to their abilities and dispositions relevant to interpreting that evidence. (p. 2)

For such hypothetical peers, we can examine the epistemic consequences of disagreements: when will it be appropriate for either party to change beliefs as a result of an argument by the other? We would hope that real reasoners would be equally persuaded to rethink their beliefs after seeing what has persuaded the ideal peers.

It will be expected that they cannot continue to maintain contrary positions, because analysis of the disagreement will also rely on the “uniqueness thesis”: “a given body of evidence justifies exactly one attitude toward any particular proposition” (Feldman and Warfield 2010, p. 6). This kind of analysis would be useful, for example, to a judge hearing a case in which multiple experts have been called to testify on both sides, and they clearly disagree. The idealization removes any concern that the disagreement might be due only to bias, stubbornness, or any other contingent property of real arguers. The argument should be completely independent of the arguers: the logic is the same, whoever says it.

An important split occurs here. On one side, we have actual arguments with real and fallible arguers. On the other side, we have arguments with ideal arguers operating under ideal conditions. The actual arguers may fail to find a relevant piece of evidence, or may fail to see a logical connection, and if so may fail to resolve their disagreement even though it could in principle be resolved. The ideal arguers will have access to all relevant evidence and all logical connections. If these ideal arguers are nevertheless unable to resolve their disagreement, that disagreement is a “deep disagreement”: there are no further resources that are accessible even in principle to help decide the matter. This split indicates what is crucial to the concept of “deep disagreement”: a disagreement for which we do not even in principle have rational resources available to determine what position is correct.

However, what this split does not indicate is how we should handle a disagreement that is “deep”. If there are no accessible resources—no additional reasons, no common ground in beliefs—that can resolve the dispute, what are we to do? It’s tempting to assume that we must simply give up on seeking a resolution. After all, that is what these disagreements are: limits on when reason can succeed in achieving a resolution. Nevertheless, a significant portion of the debate about deep disagreements is devoted to whether or not we can legitimately stop trying to work on them.

There are two approaches to settling the question of whether it is legitimate to stop trying. The first approach is to fine-tune when exactly a disagreement should be considered deep, by discussing which resources might still be accessible (Campolo 2013; Phillips 2008; Godden 2013; Barris 2018). This discussion offers different possible limits on continued debate, with a view to what rational resources might be available other than additional reasons or shared beliefs. The second approach is to consider our accountability in situations which involve deep disagreements: even if we have good reason to suspect the disagreement is in fact deep, there may still be situations in which we cannot justifiably stop the discussion. We may be accountable for continuing to try hard to keep talking to one another (Adams 2005; Dare 2016).

The first approach can be represented in a diagram which shows how the category of deep disagreements is larger or smaller depending on what precisely is used as the key indicator that further discussion is impossible. Figure 1 illustrates how deep disagreements are distinguished from what Fogelin (1985, p. 3) called “normal” arguments, according to the degree to which reasons, understood as rational resolution resources, are accessible to the participant reasoners (cf. Godden and Brenner 2010, p. 43 ff.).

Fig. 1
figure 1

Deep disagreements as a function of reason

Full size image

On the left side of the horizontal axis, reasons are highly accessible, meaning that real arguers will be able to identify and use them. Reasons become increasingly inaccessible as one moves rightward along the axis, meaning that real arguers start to have difficulty recognizing reasons or understanding their impact on the disagreement. However, in principle (if the arguers are ideal peers) reasons could still be recognized and used effectively. Moving further rightward along the X-axis, as resources become even less accessible, disagreements become increasingly resolution resistant until such a point as they become “deep”—rationally unresolvable due to the unavailability of sufficient rational resolution resources. At this point, even the ideal peers must concede defeat: there are no additional rational resolution resources available to them.

Since so much is at stake in deciding when a disagreement might be beyond rational resolution, the first line of approach is to make sure the line between deep, rationally irresolvable disagreements and normal (even if recalcitrant) disagreements has been correctly marked. The vertical dotted lines indicate spots where different theorists have proposed to draw the line between accessible (decidable) arguments and deep (undecidable) arguments. As indicated in the diagram, Campolo (2013) would draw this line to the left of Feldman (2005) because Campolo contends that it is not merely unproductive but harmful to pursue a debate across a gap of understanding: it degrades our reasoning skills, which decrease incrementally as we struggle harder to be understood. Feldman, by contrast, contends that such disagreements call for a suspension of judgement by each party, which counts as a rational resolution. Phillips (2008) would draw the line further to the right of Feldman, because she contends that we can reason across significant differences in beliefs as long as we are sufficiently alert to the procedural barriers we might experience.

The diagram shows that non-ideal conditions, where relevant reasons are not readily accessible, are left behind before we enter the range of “deep” disagreements. This is the aspect of the diagram that re-opens the question of what our responsibility is when we are engaged in a disagreement that might not have rational resolution resources. Initially, when the reasons are not readily accessible, it seems advisable to keep trying. Flawed though the reasoners may be, cognitive psychology does support the claim that they could spot reasoning gaps or mistakes and work together to find a way into better reasoning. (See Kahnemann 2011, for a good overview of the psychological factors that inhibit or facilitate logical reasoning.) However, by hypothesis, when the reasons are not accessible, further attempts are at best ineffectual and at worst counterproductive, as indicated above (Campolo 2013).

For this reason, I want to investigate more closely the second approach, which can proceed independently of the first. We already know that in addition to the deep disagreements—whichever they turn out to be—we have disagreements which may be considered “difficult”, “recalcitrant” or “intractable” because real reasoners have difficulty making full or accurate use of the rational resolution resources which are available to them. These disagreements will include ones in which there is no easy road to resolution, yet there is no good justification for giving up and “agreeing to disagree”. Medical treatment disagreements are in this category, and are the key example in Adams’ (2005) argument that we must continue to debate even when a disagreement is deep. Medical treatment decisions require consent and also require sound medical advice representing due diligence in weighing available medical evidence and the particular patient’s condition. Medical decisions also rely on real arguers: real patients and their representatives, real doctors and medical researchers, reasoning in circumstances where time and options are limited. The inability to turn to ideal peers in these situations highlights a problem with relying on ideal peers in any situations.

The assumption made in calling for ideal peers is that flaws in the argument process due to affective and social factors, and some procedural factors, can be handled independently of the reasoning process. Sometimes these flaws can be handled with existing dispute resolution options. For example, if one person dislikes or distrusts another, we could provide a mediator to facilitate communication. If one person considers a procedure too adversarial, we could train the person to be more comfortable in the procedure, or modify the procedure so that its most adversarial aspects are removed. If the original participants in an argument find themselves unable to interact with each other at all, they could transfer the substance of the argument to neutral parties who are not emotionally involved and will be able to reason successfully—as we do when a case goes before a court, trained representatives argue each side on behalf of the original participants, and a judge makes an objective decision. However, the legal system is still far short of the ideal-peer system envisaged by Feldman and Warfield. Ideal peers never make mistakes and operate in an ideal social context.

It is hard to imagine that an ideal social context does exist or can be created when we want to be sure a disagreement really is deep. As van Eemeren and Grootendorst (1988) recognized, the external factors, the psycho-social reality which sets the “third order” conditions, can be challenging to deal with. If each reasoner is to have a real voice in the debate,

The third order conditions make it clear that for argumentation theoreticians there is also an important indirect political responsibility in striving for individual freedom, non-violence, intellectual pluralism, and institutionalized safeguards for rights and means to obtain information and pass criticism. (pp. 287–288)

If social customs and pressures resist freedom of expression, or questioning authority, we will not be able to begin a productive argument because some participants whose views are needed will either not feel safe participating or will not be allowed to participate. By definition, if the third order conditions are not met, the people who are “epistemic peers” in Feldman and Warfield’s (2010) sense will lack important evidence but will have no social process to include the people who have that evidence. Social change would have to precede any change in reasoning procedures.

The sheer difficulty of changing social patterns to enable more reasoning about divergent opinions can seem overwhelming. This makes it tempting to stay with an idealized picture and make the question of everyone having a “real voice”, and real responsibilities, a separate issue. However, compartmentalization of disagreements into those with ideal social conditions and those without creates two problems. First, if a disagreement fails to include the voices and information of people whose experience is relevant to the reasoning, it is not clear we have correctly understood the substance of the disagreement. Second, we may misidentify what is required of ideal reasoners because expectations of reasoning can be biased by gender, education, and culture.

Both concerns have been raised before. My aim here is to link them to each other and to the question of when to continue discussion if a disagreement may be “deep”. Both concerns raise the “third order” pragma-dialectic consideration: the need for a “psycho-social reality” which is both inclusive and non-partisan. This means that instead of defining “ideal peers” as operating in an ideal psycho-social reality, we need to take a more pragmatic look at how third-order requirements for intellectual pluralism and individual freedom could be made more accessible in real disagreements.

One way to approach this is by introducing a second dimension to disagreement, to cover not just procedural but affective and social aspects side-stepped when imagining disagreements happening between “ideal peers”. My candidate for this second dimension is a multi-faceted description of trust which can clarify the affective and social attitudes within which reasoning takes place. The introduction of trust provides a realistic basis for assessing responsibility for continuing or ending discussion in cases of deep disagreement.

2 Deep Disagreement Between People Who are not Equals

The equality of ideal arguers is extremely difficult to achieve in practice. Existing criticisms of argument practice show that however open-minded arguers think they are, they might not be equal participants in argument. The source of their inequality is beyond their individual control. For example, consider a female in a male-dominated society, or an indigenous person who is profoundly uncomfortable with adversarial forms of discourse in a courtroom. No matter how open-minded the men or the court might be, they do not have enough control to make the woman or the indigenous person comfortable with them.

Rooney (2010), Hundleby and Rooney (2006), Hundleby (2010, 2013), Lang (2010) and others point out that adversarial argument often disadvantages women and members of any culture who do not typically engage in combative and competitive practices. Hundleby points out in “Aggression, Politeness, and Abstract Adversaries” (2013) that even the most respectful adversarial argument is still a method of probing and challenging the other, and this probing and challenging upholds a standard which favours the social practices of Euro-American white males.

What can happen when one arguer is expected to conform to reasoning practices and behaviour that are not comfortable or safe for them is that their evidence and reasons are discounted. They are, in Govier’s (1993) term, “rhetorically disadvantaged”: they are not considered credible. They are treated in ways Fricker (2009) characterises as “epistemic injustice”: their evidence is discounted in ways that cannot possibly be justified on moral grounds and therefore cannot be justified epistemically. Either they are not believed simply because of who they are (for example: indigenous, female, too young), or they are ignored because the society has no concept to cover their experience (for example: child abuse, sexual harassment, or post-traumatic stress disorder). They are undermined as knowers, and Fricker says

The undermining of someone as a knower is, conceptually and historically, closely related to their being undermined as a practical reasoner. (p. 137)

If a reasoning process places some reasoners at a disadvantage due to their social position, then, as indicated earlier, the reasoning which identifies a “deep disagreement” may be seriously misinformed or inaccurate. The root of the disagreement may not be the subject matter in open debate; there may be hidden elements relating to the lived experience, emotions, and priorities of the arguers. For example, Friemann (2005) and Gilbert (1994, 1997) have argued that excluding emotion and its impact on arguers can create a significant misunderstanding as to what the disagreement really is. In “Emotional Backing and the Feeling of Deep Disagreement” (2005), Friemann uses Gilbert’s (1997, pp. 126–127) example of how reasoning shifts from one set of premises to another as emotion becomes safe to reveal. What seemed to be a disagreement about the importance of tradition in preserving a newsletter’s name turns out to be a far more easily resolved debate once it is clear that one arguer is concerned less about tradition than about a potentially overwhelming task. This concern can be dealt with successfully once empathy enters the picture and practical suggestions for workload can be introduced.

The related concern is that we will have some real, and deep, disagreements that cannot be given to anyone other than the original participants in the disagreement. We cannot refer these disagreements either to real but impartial reasoners such as arbitrators or judges, or to ideal peers who have perfect reasoning because only the people actually involved can justifiably participate (for example, Adams 2005, to be discussed later). The two concerns are linked. If the standards for ideal reasoners cannot be met by the right variety of real reasoners, then those standards themselves may need to be the subject of social change.

This is when the two approaches to identifying and dealing with deep disagreements overlap. It is only an assumption that the ideal peers who can judge when a disagreement is deep will bring to their deliberation a fully inclusive standard of evaluation. If the current standard of objective evaluation is as deeply rooted as it has been in practices which favour educated Euro-American males, we can’t yet be sure we know what a more inclusive standard would look like. We can’t be sure that the identification of deep disagreements can proceed independently of our ability to deal with the pragma-dialectic “third order” of ensuring a fully inclusive social context for the reasoning.

Existing suggestions for handling deep disagreement are consistent with acknowledging that we are aware of the importance of the third tier of pragma-dialectics and do have at least an “indirect political responsibility” (van Eemeren and Grootendorst 1988, p. 287) to make it possible to engage in further reasoned argument. Campolo (2013) suggests turning to a building of greater understanding in and across disciplines. Godden and Brenner (2010, p. 57 ff.) suggest the use of “rational persuasion,” which they explain as “sort of rhetoric in the service of concept-formation.” Friemann (2005) argues that if emotions are an integral part of logical thinking, then psychological methods of dealing with conflict are relevant to handling deep disagreements. Dare (2016) turns to government action that can be justified as a reasonable part of putting public policy into practice. Adams (2005) argues for continuing to reason, as a matter of social responsibility, without any significant change in procedure when the disagreement is deep.

Perhaps because of the focus on ideal arguers who have already been established as equal, the emphasis has been on the responsibility that can be taken by individual arguers. For example, Dana Phillips, in “Investigating the Shared Background Required for Argument: A Critique of Fogelin’s Thesis on Deep Disagreement” (2008), argues for individual responsibility in overcoming obstacles to reason. Phillips wants to demonstrate that productive argument is possible even between arguers who lack shared beliefs because of cultural distance (p. 87). She argues that development of “shared procedural commitments” not only constructs a way to argue together but permits the argument to handle otherwise deep differences in participants’ core beliefs (p. 86). The procedural commitments become the “second-order” requirements of pragma-dialectics: the “attitudes of the discussants” (p. 97), which Phillips sees as something over which the arguers can have control:

procedural barriers to argument tend to surface in contexts where such shared beliefs are sparse. Being conscious of those barriers and working to overcome them where possible is perhaps the most crucial step to enhancing the power of argument in our world. (p. 101)

Phillips’s focus “is on what arguers need to have in common in order to argue and not what the external circumstances of their discussion must be,” so she explicitly disregards third-order conditions (pp. 97–98). I take Phillips to be creating an idealization similar to Feldman and Warfield (2010). We assume background conditions are the same for all, and if argument is possible in those conditions, we focus on the attributes an individual must have to argue productively across cultural distance.

I agree with Phillips’s contention that productive argument could happen across significant cultural differences. However, Phillips also appears to assume that because belief differences can in principle be handled with a shared commitment to procedure, it is not necessary to deal with any affective or social obstacles to argument as long as we are “conscious of those barriers and working to overcome them.” The difficulty is that consciousness and an ability to work to overcome barriers may not be simply a matter of personal disposition, intellectual assent to equality, or reasoning ability. As Fricker’s (2009) many examples illustrate, it is easy for even well-meaning people to be blind to the social prejudices which have already skewed their judgment.

Barris (2018) disagrees with all the suggestions above, because he contends that deep disagreements cannot meaningfully be handled by any generic reasoning strategy, not even a generic open-mindedness. Barris’s key contention is that deep disagreements are incommensurable, in that the rational resources brought to each party’s understanding of the dispute are mutually incomprehensible.

Even scholars who acknowledge deep diversity as an important category of disagreement and insist on our openness to such deep differences typically overlook the profound complications involved in our coming to understand these differences in the first place. (p. 392)

Nevertheless, Barris offers an option that he argues is still within the realm of rational resources, and still available to the individual participant. He suggests we should use the recognition that we do not understand the other side to alert us to the need to make an “existential decision”

By existential decisions I mean decisions that express the particularity of who we are, and in that sense express our being. …These decisions are personal not in the sense that they depend on arbitrary whim, but in the sense that the responsibility for making them cannot be handed over to someone else, but ultimately rests on the person deciding, and on that person alone. (p. 399)

This bridges the two questions of how to identify a deep disagreement and when to take responsibility for continuing to debate, but it does so at the same “second-order” level of procedure used by Phillips. Barris does not directly address whether his solution offers a “third-order” assurance that all participants are equally respected, only that no position is rejected because of difficulty understanding it.

Govier’s extensive work provides a more complex mix of individual and social responsibility. In her work on argumentation theory, she presents reason as a practice that can help us work collectively through difficult debates provided we are respectful of one another and attentive to one another’s arguments (Govier 2001, p. 3). For Govier, reasoning is part of an ongoing exploration of how good citizens can deal fairly with one another. If we are fair, respectful, and reasonable, then even in adversarial argument we will create a “safe” space within which we can discuss any topic at any length, even in the absence of trust between the arguers (Kloster 2016, p. 66). This is primarily an individual responsibility. Her work on reasoning makes relatively little use of her work on social responsibility, although she is consistent across both in seeking to build community and to use reason in doing so (Kloster 2016, p. 68). She discusses social trust in Social Trust and Human  Communities (1997), and individual trust in Dilemmas of Trust (1998).

The consistency in Govier’s approach both to reasoning and to social conflict suggests we could take the next step, and tackle both the social and the individual obstacles to reason. In her work on social conflict (1997) she concludes that

To reach important goals, we must act, and to act we must preserve a sense that what we do can make a difference (p. 254)

In hoping that we will act rightly [on collective concerns] I must hope not only that other agents do so but also that I myself can do so. (p. 256; italics in original).

If deep disagreements are on matters of general social concern, such as the right to life, or freedom of speech, then the reasoners who study them have more than just an “important indirect political responsibility in striving for individual freedom, non-violence, intellectual pluralism, and institutionalized safeguards for rights” (van Eemeren and Grootendorst 1988, p. 287). They have a positive opportunity to use an analysis of deep disagreements as an indicator of where we can contribute to social change. This is a wider responsibility than Barris’s suggestion of individually making a responsible “existential decision”.

The opportunity to be instrumental in wider social change brings together Govier’s work on political struggle with work by feminists on challenging the social context of an argument whenever the “ideal” arguer will be unaware of or unable to deal with systemic bias. For example, Hundleby and Rooney (2006) express the concern that treating reasoning as separate from politics blocks the opportunity to recognize diverse and more equitable perspectives.

In challenging the traditional philosophical segregation of reason and knowledge from politics, feminist and liberatory epistemologists …argue …that understandings of reason and knowledge need to engage more constructively with the ethical and social specificities that frame …social and political knowledge projects that explicitly seek to advance social justice. In particular, such understandings draw attention to the fact that the ways in which theorists conceptualize, think, or reason about social and political issues have regularly given voice to specific perspectives over others, thus limiting opportunities for insight and resolution. (p. 2)

Following Govier, if we recognize that procedural questions about reasoning already overlap with social practice, we can also look for approaches that deal with disagreement at a more interpersonal and systemic level. If we are less concerned with identifying deep disagreements and more concerned with our responsibility to continue or end a “deep” or “recalcitrant” disagreement, we can continue to examine procedures in a broad sense. Procedures are not just those which govern the use of rational resolution resources, but also those which guide the behaviour needed for demonstrating an appropriate level of responsibility. Trust by individuals in one another and trust by individuals in their social procedures and institutions are two categories of trust that offer a bridge between individual and social aspects of argumentation, between reasoning capacity and responsibility.

3 Trust as a Measure of Affective and Social Obstacles to Reason

If reasoners do not feel safe, or respected, they do not speak. In argument, we know that the third-order conditions have not been met because each voice is not heard. Safety can be seen as a function of the trustworthiness of the other participants and the procedure: I am safe if I can trust the others not to proceed in a way which ignores or harms me. Trust is an umbrella concept that can cover both the affective feeling that I am safe here and the rational calculation that I can trust these people because they have too much to lose if they harm me.

There is already considerable literature on trust as a way of interpreting people’s interactions in contexts in which arguments may be given and assessed, such as in business decision-making and organization management, and there is general agreement that trust is a necessary factor in social functioning. If the umbrella is extended to the social level, trust can be multi-level, “individual, group, firm, and institutional”, and have multiple roles, “cause, outcome, and moderator” (Rousseau et al. 1998, p. 393). This permits trust to be seen as a factor in how social expectations and practices govern whether emotions can be expressed in argumentation, which interpersonal relationships should be sustained in spite of disagreement, which argumentation processes can be tried in which contexts, and which disagreements can be brought into the open for discussion. All these social elements play a role in understanding procedural elements of disagreement.

The definition of trust and its applicability to the understanding of interpersonal or organizational behaviour is not settled. In “Understanding the Trust-Control Nexus” (2005), Bijlsma-Frankema and Costa acknowledge that even after earlier attempts to integrate different lines of research (e.g. Rousseau et al. 1998), “different definitions of trust abound in the literature and agreement on an exact definition of trust is seldom found.” However, there are consistent elements such as “positive expectations and the willingness to become vulnerable” (p. 261).

One advantage in connecting the study of trust to how well individuals can reason with one another as equals is that the traits or dispositions of people who trust coincide well with the traits and dispositions of a good reasoner. Reporting the benefits of trust discovered by the studies they examined, Biljsma-Frankema and Costa (2005) list: “open communication and information exchange”, “psychological safety”, “commitment”, “belief in information and acceptance of influence”, “mutual learning”, “attribution of positive motives” and “positive outcomes such as high levels of cooperation and performance” (p. 263). More directly, Rousseau et al. (1998) report that trust “has long been found to be an important predictor of successful negotiations and conflict management efforts” (p. 396, citing Deutsch 1958). In effect, the literature on trust offers assistance in where to look for help in building the open-mindedness and open disclosure of information that were mentioned earlier as ways for individuals to approach deep disagreements. It also opens up a way to consider the interpersonal, social aspects of procedures that are important to Govier and to a “third order” approach in general.

To resolve a conflict, arguers must feel safe disclosing relevant information. For example, in Friemann’s (2005) discussion of Gilbert’s (1997) example, it seems that no matter how open-minded one arguer is, the real point of disagreement will not be revealed by the other so long as she feels unsafe disclosing personal weakness. Preston-Roedder (2017) treats this sense of safety as something which is not merely interpersonal but is reflective of broader social harmony that facilitates effective interactions in general.

I argue that one condition someone must satisfy in order to live in such harmony with people is being worthy of their civic trust; roughly, she must behave in ways that make it appropriate for others to interact with her without fear. (p. 4)

I will take advantage of one categorization in the existing literature in which trust is considered as having multiple aspects. In the literature they review, Bijlsma-Frankema and Costa include a categorization that maps well onto the factors that seem to cover both affective and social dimensions of argumentation as well as the full range of procedural issues:

Zucker (1986) distinguishes three types of trust: (1) character-based trust, based on social similarities and shared moral codes; (2) process-based trust, based on experiences of reciprocity; (3) institution-based trust, flowing from institutional arrangements that evoke and sustain trustworthy behaviours. (2005, p. 261)

This categorization presents trust not merely as an individual choice to risk believing or dealing with another person, but also as a measure of ability to function effectively in an organizational or social setting in which there is no prior personal relationship between the parties. For my overall purpose it is not crucial that this definition of trust is the best choice, only that it allows us to set some guidelines for when a disagreement may founder because social practices that set the context for reasoning are themselves in need of change.

Zucker’s categorization of trust as described by Bijlsma-Frankema and Costa can be connected with affective, social, and procedural aspects of disagreements in the following way. “Character-based trust” connects with affective issues in argument in that greater recognition of social similarities, shared values, and predictable behaviour by a person makes it easier to feel safe enough in that person’s presence to engage in reasoning, exhibit emotion and make personal disclosures. The positive influence of feeling safe is the factor Friemann (2005) thought opened up the possibility of resolution.

“Process-based trust”, based on “experiences of reciprocity”, fits very well with the process-based considerations discussed by Phillips (2008), described as conforming to the second-order considerations of pragma-dialectics. This is the type of trust which generates procedural commitments to positive behaviour or results from trustworthy behaviour within the procedure, and is the type of reasoning procedure that enhances the ability to follow patterns of introducing new ideas or weighing evidence.

Trust in procedures at the level of being willing to use a structured procedure, such as negotiation, may fit better as “institution-based trust”, which connects to the social aspects of argumentation. It is “institutional” arrangements such as company practice, community norms, or government and court provisions which provide the background stability for strangers to interact, or for a person who has been discriminated against to protest the lack of equal treatment.

Trust is not a static condition: it can emerge, grow, drop, or disappear. Since “risk and interdependence are necessary conditions for trust, variations in these factors over the course of a relationship between parties can alter both the level and potentially the form that trust takes” (Rousseau et al. 1998, p. 395). If the studies of trust show when and how different aspects of trust evolve over the course of interactions, we have a source for interpreting why people are not reasoning well together and what changes might be needed to restore productive argument. This will help us decide when it is our responsibility to continue to debate even if a disagreement is deep, or cease debate even if a disagreement is not deep.

However, I do not want to close off the possibility that “deep disagreements” could still in some cases be dealt with in isolation, as matters only of argumentative method and choice of premises. I will integrate a trust dimension in a format that permits reasoning (beliefs and logical procedures) and affective and social elements, including social procedures, to be considered separately.

If disagreements can sometimes be handled by rational resources alone (by “ideal peers”) and yet sometimes require fallible reasoners in a non-ideal social context, then the limits on the use of reason as the means to resolve a disagreement are a combination of two factors. One is accessibility to reason; the other is trust, measured as a combination of the character-based, process-based, and institutional trust. Introducing trust as a component of disagreement is intended to capture the affective, procedural, and social factors which influence participants’ willingness and ability to engage in reasoning.

Here is how the difference between deep disagreements and others could be presented as a two-dimensional picture, with access to reason along one axis and level of trust on the other.

In Fig. 2, the horizontal axis, depicting the relative accessibility of relevant reasons to the disagreement, is repeated from Fig. 1. The vertical “Trust” axis depicts the amount of trust, summed over the different varieties (character-based, process-based, and institution-based) operative between discussants in the situation—i.e., the overall level of trust in the individuals involved, the reasoning process used, and the social context in which the dispute arises. The white areas in the figure represent the set of disagreements that are amenable to rational resolution. The grey-shaded areas in Fig. 2 represent the set of disagreements that are “recalcitrant”, intractable, or rationally unresolvable, due to the combination of factors of accessibility of relevant reasons and the amount of trust in the situation. The line dividing these regions indicates the amount, y, of trust required to rationally resolve a disagreement with x amount of accessibility to rational resolution resources. Structurally deep disagreements are depicted on the far right of the figure as being rationally irresolvable irrespective of the amount of trust in the situation.

Fig. 2
figure 2

Rational resolvability as a function of reason accessibility and trust

Full size image

In Fig. 2 accessibility to reason is dependent on argument taking place in a context in which all reasoners have sufficient trust to be able to continue to argue with one another, no matter how far apart their views seem. The less trust they have in one another and the process, the more quickly they will discover obstacles to continuing their argument. The more trust they have, the more easily they will overcome potential obstacles and continue their discussion.

The standard view of deep disagreement as due to purely structural features of the disagreement (namely the accessibility of relevant reasons) and not due to features of arguers like temperament is not lost, only expanded. Moving across the horizontal axis itself (i.e., considering disagreements where there is minimal trust between participants), arguers can reason, trusting only their own logical acuity, until there is either a resolution, or they reach an impasse—a deep disagreement. The right side is the space in which a disagreement is deep no matter how much trust there is between parties: even idealized arguers with a shared social context cannot resolve their disagreement. As mentioned in discussing Fig. 1, the line separating deep disagreements from those that might be resolution resistant but are not structurally intractable might vary depending on one’s theory of depth or intractability.

Without considerations of trust, we would have just these two blocks—one in which disagreements are structurally rationally resolvable and the other in which they are not. That was the picture presented in Fig. 1. What the trust dimension adds to the diagram is a middle section in which disagreement might or might not be accessible to reason. We now also have the ability to assess other deadlocked or potentially undecidable disagreements. The shaded region labelled “recalcitrant” (rationally unresolvable) disagreements” indicates disagreements that might not seem accessible to reason, but cannot yet be shown to be deep disagreements. The diagonal separation of shaded from unshaded regions in the diagram indicates that more trust makes it easier to stay in an argument. Where there is maximum trust, arguers trust one another to be consistent, fair, and respectful, trust the process they use to deliver clear, accurate reasoning, and trust the social context to keep them safe so they lose nothing by engaging in argument on either side. With this level of trust, they are on an equal social footing, and if they cannot resolve their disagreement, they can conclude they are in a deep disagreement, though they must still consider the question of whether they ought to end or continue their debate. In less than ideal cases, one or more types of trust may be low or missing completely. For these disagreements, accessibility to reason will be a function of whether the arguers trust each other, or at least trust the argumentative procedures, or, if they don’t trust one another or the process, then at least they trust the background institutions of their society—their government, courts and business organizations—to provide enough stability that they can argue without significantly jeopardizing their safety. The less trust they have, the more likely it will be that they cannot reason with one another, but we will not yet be able to conclude they have a deep disagreement. The boundary might be less linear and much fuzzier than it has been drawn here, but all we require is that trust can be a factor in making arguments inaccessible to reason.

Trust also provides a partial solution to deciding how to handle a difficult disagreement, not just an analysis: it permits us to consider the “procedural” aspect of a deep disagreement in broader terms that open up more potential remedies. It also permits us to identify legitimate ways to proceed when the substance of the dispute cannot be dealt with by ideal arguers, in the timelines required or by the individuals who are required to participate. Having legitimate ways to proceed when there is not enough trust is one way to handle questions of responsibility. We are responsible when we ensure no unjustifiable damage is done to relationships or social systems. Therefore, we can continue to discuss a disagreement whenever we can maintain or rebuild trust; we need to halt a discussion when trust has eroded to the point where participants feel unheard or unfairly dealt with.

Exploring a connection between deep disagreement and trust may be mutually beneficial. The literature on trust concentrates on what actions, such as co-operation, increase trust, and what actions, such as betrayal, destroy it. In Trust, Distrust and Trustworthiness in Argumentation: Virtues and Fallacies (2013), McMurphy suggests that argumentation theorists can contribute to understanding trust, by considering how trust might be a factor in both successful and unsuccessful argumentation. One key question is whether the preservation of trust should be a goal of argumentation, and whether, when assessing argument quality, we should take into account any potential impact on the well-being of a trusting participant. I follow McMurphy in taking these questions to be important to a full understanding of deep disagreement, because maintaining or increasing trust is a social factor connected to the equality of voice we are hoping for in productive argument.

4 Trust, Reason, and Social Change

Using trust to account for some of the obstacles to reason gives us one way to reach towards social change, because if the problem originates with either procedural trust or institutional trust, we can look in that direction to build trust as a way to create greater equality. As Lang (2010) points out, asymmetrical power is not easily rebalanced for equality.

The specificities of asymmetries can be complex, involving age, gender, race, culture, language, and institutional roles...The degrees and types of harm [caused by asymmetrical power relations] can vary dramatically…. (Lang 2010, p. 323)

Social structures and expectations, not personal preferences, dictate when a discussion may happen and who may participate. If power is in the right hands and is being exercised responsibly, there should be trust in the process and the institution within which it is used. In addition, where we trust legitimate authorities, we may also trust them to make decisions—even, perhaps, in the case of inadequate time for reasoning or undecidable issues (as will become clear in later discussion). If power is exercised responsibly, that means we trust that the people with more power than ourselves have our best interests genuinely at heart in their deliberations. We also trust that those with less power than ourselves are not leaning too hard on us and abdicating their responsibility to inform us of their concerns and interests to guide our discussion.

If there is insufficient trust in adversarial reasoning processes, for example, we can investigate which other processes are trusted. There are a number of options for non-adversarial argument, including “coalescent” argument (Gilbert 1997, p. 106), in which both participants are engaged primarily in an open exploration of possibilities as opposed to an attempt to persuade the other participant of the superior merits of their own pre-established position. A “talking circle”, based on indigenous practice, builds in equality by having each person speak only in turn and speak without interruption. Facilitation or mediation using a neutral party is a third option, which only works if there is both character-based trust in the facilitator/mediator and process-based trust in facilitation or mediation to produce a better outcome than litigation. Fricker (2009) suggests familiarization as an option to increase credibility, hoping prejudice might correct itself over time:

…with the degrees of familiarity – gained over the duration of a conversation or perhaps a more sustained acquaintance – the prejudiced first impression melts away and the hearer’s credibility judgment corrects itself spontaneously. (p. 96)

All forms of collaboration, exploration, and consensus-building need at least enough trust in the process to be sure all parties will feel safe enough to participate. With a trust-based interpretation of the social context in which a disagreement takes place, we can assess some examples of disputes to see what it takes for them to be accessible to arguers (real or ideal) and whether there are processes which allow us to restore productive discussion even when circumstances are not ideal.

5 Examples of Multi-faceted Intractable Disagreements

All the following are situations which raise questions about whether reasoned discussion could justifiably be terminated without resolution of the disagreement. Some might comfortably fit Fogelin’s definition of deep disagreement as summarized by Godden (2013). However, some sit in a no-man’s-land where it is not clear what has made the situation reach stalemate and whether reasoning can justifiably be terminated because of the stalemate. Each of them will illustrate how trust can clarify the nature and extent of the disagreement, and in some cases offer new ways to proceed.

5.1 Scenario 1

This situation occurred unexpectedly in happy conversation with a friend at a café. Our philosophical backgrounds have minimal overlap; we work to understand one another’s reasoning. On this occasion, the topic was identity politics, on which we disagree. I had just begun to probe her viewpoint when she stopped the conversation, saying, “I feel interrogated.” She had become profoundly uncomfortable. I immediately became profoundly uncomfortable as well, fearing I had shattered our friendship.

I include this example because my friend and I should count as epistemic peers, equally able and informed, who have had many discussions on topics like this before—but not this topic. I intended my questions to clarify her views, but she experienced the questions as hostile or threatening. Somewhere in the topic or the process was a problem neither of us anticipated. Should we have stopped the conversation and if so, why? Were there background issues not adequately addressed as we entered the discussion, or was this a deep disagreement at the level of core beliefs and reasoning process?

5.2 Scenario 2

In this situation, I was the one made uncomfortable. In a department meeting, a colleague raised a question about whether to stop using examples about abortion in the critical thinking class, after one student complained. Discussion of abortion made the student uncomfortable. Four colleagues felt strongly that classroom discussions in philosophy ought to make students uncomfortable. I was alone in arguing for adapting examples to suit students’ concerns. Within minutes, I no longer felt I had enough credibility to continue to defend my position. I became so uncomfortable I wanted to leave, yet I had become too emotional to handle the social niceties of excusing myself. What seemed to begin as a coalescent exploration of possible positions had become an adversarial championing of pre-existing positions, and I was no less adversarial than the others.

In this example all participants reason well and should be comfortable with the discussion process, but discussion was unproductive. In addition, the topic of discussion was students’ ability to deal with classroom topics which might trigger discomfort, yet no students were present for the discussion. This raises the additional question of whether, by limiting the discussion to faculty, we had pre-empted their voices and misjudged aspects of their concerns, which in turn raises questions not just of the depth of disagreement but also of the responsibility to ensure the “intellectual pluralism” which is part of our responsibility in attending to the social context of argument (van Eemeren and Grootendorst, pp. 287–288).

5.3 Scenario 3

There was a dispute between business owners and an entrepreneur preparing to start a consulting firm in their city. The entrepreneur lives with a disability, and his proposed business was to increase opportunities for people with disabilities to network and find employment on an equal footing with everyone else. Local business owners had earlier agreed to participate in one of his projects to increase local employment of people with disabilities, yet they initially offered the consultant money to start his consulting firm elsewhere. He refused. They changed their strategy, and met in person to tell him he would not be able to succeed in business even with financial backing. When he attempted to question their inconsistent reasoning, he was ignored.

Surprisingly, on other occasions, he experienced similar dysfunctional argument from the charities that normally advocate for people with disabilities: they argued that he was violating norms for disabled people by proposing a for-profit business instead of a charity and failing to appreciate how much the charities provided for people with disabilities.

Fortunately for the entrepreneur’s self-respect, he knows that the belief that disabled people are necessarily mentally incompetent and must depend on and be grateful to others is a historical construct, traceable to the Industrial Revolution’s need for able-bodied workers and to the eugenics movement’s belief that deformity of mind or body was a genetically-transmitted defect in the human species (Snyder and Mitchell 2006).

I include this example because it would have been a deep disagreement through the 1970s, until competent arguers resolved the disagreement in favour of equal rights for people with disabilities. Since 1985, Canada’s Charter of Human Rights and Freedoms, s.15, has prohibited discrimination on the grounds of disability. Yet the Charter, reflecting agreement by good reasoners, has not changed majority attitudes. The extreme resistance by both business owners and charities shows that for them this still feels like a deep disagreement. From their perspective, the reasoning used to amend the Charter remains incomprehensible. In effect, if this is not a deep disagreement at a social level, it is still “recalcitrant” at an individual or community level, and there is no easy justification for simply ending the debate.

This example is comparable to Fricker’s (2009) many examples of epistemic injustice. Here, the injustice is to treat people with disabilities as a group who do not deserve to be listened to. The entrepreneur is a competent and open-minded arguer, but does not have enough power to require the business owners or charities to be more open-minded in listening to him. However, any attempt to replace him by another arguer his opponents would respect is another form of discrimination. It is his voice which needs to be heard, and this is a third-level problem in pragma-dialectics: social change has not come far enough for him to make himself heard. Should he give up at least for now? Must his be the voice that strives to break through their resistance to reason? Whose responsibility is it to work on this disagreement that cannot fairly be postponed without harm to people with disabilities and the community as a whole?

5.4 Scenario 4

Adams (2005) presents the difficulties of reaching an ethical consensus on medical treatment. In the Schiavo case, Schiavo was comatose and her parents and husband disagreed on treatment options. As Adams (2005) describes the nature of the disagreement,

there was here a lack of agreement both with respect to substantive beliefs (Can she feel pain?) and also with regard to procedural standards (How can we settle whether she is feeling pain?). In this way, the division over Schiavo’s neurological status looks like a deep disagreement, and thus not one amenable to a reasoned argumentative resolution. (p. 71)

The disagreement meets the criteria for a deep disagreement. The reasons to continue or cease life support have not reached an unequivocal conclusion. However, Adams resists categorizing it as beyond the reach of reason, because giving up on reasoning in such cases abandons responsibility just when it is most important:

a large part of what makes most bioethical dilemmas in clinical care so wrenching for those involved is the necessity to make decisions under conditions of uncertainty, and this because all of the possible courses of action open to a physician or family in such a case (for example, to start a patient on a ventilator now or to wait to do so) impose costs – emotional, physical, and financial. Hence the moral pressure to make a decision and accept the consequences, whatever they may be. (p. 76)

I include this example to show why there is a difference between deciding when something is a deep disagreement and deciding whether we have a responsibility to continue discussion of the disagreement. As Adams argues, participants in cases like this may be responsible for continuing discussion as long as time permits even if further discussion will not resolve the disagreement. It also illustrates the problem with hoping the disagreement can be referred to ideal arguers. We must include certain voices in the reasoning even if they are not ideal arguers. Here, whatever epistemic disadvantage or prejudices they might have, it is the family members themselves who must debate whether to continue life support. They have a legal responsibility to their kin, and the medical system has a legal obligation to obtain their consent when an individual is comatose. Citing Core Competencies for Health Care Ethics Consultations (1998), Adams notes that the aim of health care ethics consultation is to construct a “consensus among involved parties” described by a range of “morally acceptable options” within the context of a particular clinical case (pp. 66–67, italics in original). In such situations, the people who disagree are expected, and often required by law, to consult or be directed by an institutional health care ethics committee, so there is an institutional component aimed at reducing epistemic injustice as well as ensuring a timely decision.

5.5 Scenario 5

Tim Dare, in “The Normative Significance of Deep Disagreement” (2016) presents a situation which might be a deep disagreement and is ambiguous as to who should argue and whether argument should continue. The government of New Zealand proceeded with a vaccination program in spite of deep disagreement between pro-vaccinators and anti-vaccinators. Dare argues it is reasonable to continue to vaccinate even when a portion of the population remains vehemently opposed to it. He counts this as a deep disagreement, and one which demonstrates that a deep disagreement does not prevent the use of reason by either side.

a diagnosis of deep disagreement …might have positive implications for our attitude to others and our appreciation of the reasons that motivate them. … recognizing that we are parties to a deep disagreement may make us treat them with more respect. (2016, p. 8, italics in original)

The New Zealand government’s approach to strengthening its vaccination program does not attempt in any way to coerce the people who have taken a principled stance against vaccination, but acts in accord with the majority position and continues to encourage people who have not yet summoned the energy to vaccinate or take a clear stand for or against. The government strategies (incentives, raising awareness, and anecdotal evidence) offer

some normatively and perhaps epistemically acceptable persuasive strategies, … [which] are not arguments but they neither completely bypass agents’ capacities for critical reflection nor counsel disregard for the perspectives of those with whom we disagree. (2016, p. 8)

I include this example because as in Scenario 3, the vaccination debate may not be a deep disagreement: it may have been resolved using good reasoning to accept the best scientific evidence in favour of vaccinating. However, as in Scenario 3, vocal opposition continues and the opposition indicates that they may not understand the reasoning accepted by the government. Since the government sides with the majority, it is not clear whether the voices of the minority are thereby placed at what Govier (1993) calls a rhetorical disadvantage and Fricker (2009) calls epistemic injustice. The government may not “counsel disregard for the perspectives” of anti-vaccinators, but nor does it ensure the anti-vaccination side argues from a position of equal strength or feels respected and heard.

I will use these five examples to consider how both “recalcitrant” and deep disagreements may be illuminated by identifying problems in level or type of trust.

6 How Trust Interacts with “Depth” of Disagreement and Responsibility to Debate

Using Zucker’s three-part categorization of trust introduced earlier by Biljsma-Frankema and Costa (2005), we can test these five scenarios to see whether reasoning should continue, who must be included as reasoners, and how we are working towards more inclusivity and equality in the social practices related to disagreement.

6.1 Scenario 1

The friends who disagreed were good reasoners: objective; able to detect incompatible assumptions or competing philosophical principles, respectful of one another’s points and able to adjust their attitude towards a proposition on the basis of good evidence. We should have been able to continue to the point where the disagreement was revealed as deep because of different framework beliefs. We didn’t; we stumbled and fell long before that. The obstacles to reason were revealed during the argument, with no obvious way to prevent them in advance or leave them until later.

The sudden change during the discussion revealed a difference in trust and safety, experienced by both participants. The situation was experienced by one participant as “coalescent” in Gilbert’s sense (1997), but experienced as adversarial by the other. The friend who felt “interrogated” seemed not to trust me to stay within the bounds of friendship in our conversation. The “interrogative” questioning was evidently a procedural barrier in Phillips’ (2008) sense. This example is comparable to Friemann’s (2005) examples, where argument creates strain in maintaining an important relationship. Because the procedure caused the drop in trust, it was not a barrier that could have been worked on while continuing the disagreement.

Change in trust level not only shows what has tipped the disagreement into being recalcitrant and possibly deep, but also offers a more robust interpretation of how we might “work to overcome” procedural barriers. Trust in shared values might be discussed once emotions have calmed. An opportunity to affirm that we value the friendship, and that a change in questioning procedure would be acceptable, could have allowed us to reopen the issue. A discussion of our shared values about friendship could have restored a coalescent approach: part of the friendship is to explore intellectual issues from divergent perspectives. However, nothing in the friendship requires that all issues should be open to discussion. This is one case in which disagreement could responsibly be transferred to “ideal arguers,” in the pages of peer-reviewed journals. If a resolution of the disagreement is revealed, we both will likely understand it and adjust our attitudes accordingly. This is an example which would support McMurphy’s (2013) hypothesis that the preservation of trust should be a goal of argumentation, and the impact of adversarial questioning on a trusting participant should be assessed in determining whether the reasoning is of sufficient quality.

6.2 Scenario 2

When the colleagues disagreed, procedural commitments were shared: adversarial reasoning is the norm. However, no-one took the role of chair in the discussion so one common procedural option for controlling debate was lacking. In addition, two colleagues seldom contribute to discussion; each contributed only once on this topic before falling silent. It is possible that they have less trust in adversarial reasoning, and the only unusual feature of this discussion was that a third individual, normally comfortable with adversarial reasoning, experienced for once how uncomfortable it can be. But as in the first scenario, it is not clear this could be cleared up by discussing and agreeing on procedure in advance.

The topic of discussion was the right balance of respect between the professors’ right to choose their methods of instruction, including their choice of examples and the students’ comfort in the classroom. This is most likely a deep disagreement in beliefs about freedom of speech and whether it trumps discomfort on the part of the hearer. There was agreement that some examples might indeed cause significant emotional distress to some students. However, it was argued that distress might perhaps have a desirable impact on the students by taking them out of their comfort zones, and certainly the distress could not be sufficient reason to change or omit the contentious examples. The arguments presented in favour of using whatever examples the professors chose relied on reasoning that everyone is in fact equally free to present their point of view and students are free to speak up or leave if an example upsets them.

As we argued, I felt a sharp drop in credibility, as if my colleagues did not trust what I said to be cogent. And perhaps I was not cogent. I had spent three years working as a faculty advisor to the campus Pride collective, learning what it meant to be an ally and not a principal in a move for social change. I knew how seriously unsafe Pride members felt on our campus in a generally conservative community. Two members of the Pride collective had been the ones to raise the problem of abortion examples with my colleague. Knowing how strongly they felt about the issue, I thought I was rising to the occasion in defending their position. On further reflection, however, it is not clear that I or any other ally should speak for the principals when the topic is one in which lived experience matters to understanding. I am not upset by the examples: can I speak for those who are?

Interpersonal trust was quickly restored, but trust in the procedure was not. It is not clear that the reasoning procedure we were using is fair, because it might falsely presuppose that anyone who disagrees can and will speak up: not only the students in the classroom, but also all faculty members present at the meeting. This is one form of the epistemic injustice Fricker (2009) cautions against: silence is often misunderstood as submission or assent (Glenn 2004). Student silence in the classroom and faculty silence in a meeting are not reliable indications that reasoning is understood and accepted. Any continuation of the debate to test whether the disagreement is deep depends on checking that trust between faculty and between faculty and students is maintained. If the disagreement is only on whether each faculty member should be free to choose classroom examples or whether there should be a departmental policy, then there is probably enough interpersonal trust and trust in group debate protocol for the department to resume the discussion with some procedural improvements. If that can’t be achieved, then as in Scenario 1, discussion could be transferred to epistemic peers and their reasoning relayed to the faculty members to consider. On the other hand, if the disagreement is on whether in principle students must be prepared to face whatever examples professors choose to provide, then faculty may have a responsibility to find a way to include students as equal participants in the debate to be sure they will continue to trust faculty in the classroom. There is no easy way to transfer this version of the discussion to epistemic peers, since there are no existing structures in which students can be confident they will be heard as equals.

6.3 Scenario 3

In the third case, trust at any level is minimal. The business leaders are well aware that the consultant’s aim is to bring about social change, by having his own voice heard and increasing the inclusion of all community members in community decision-making. They prefer reasoning to be done by themselves as community leaders. His personal and academic expertise in matters of disability did not help his credibility, even when he could point to studies that show inclusion of people with disabilities is to businesses’ economic advantage.

If he is able to change the paradigm of what a disabled person can be expected to do, his success would be a challenge to the social structure in which the current business owners have higher status in the community. The perceived threat to status may be what drives them to inconsistent reasoning. They had agreed to participate in his employment project, yet argued that employing more people with disabilities would “violate norms.” They also argued that because of his disability he would be incapable of success, and that he is already far too successful even before opening his consulting business. The inconsistency in their actions, if not their reasoning, can be explained if they are willing to hire people with disabilities into minimum-wage jobs but do not want to see educated people with disabilities expecting a say in how the town and its businesses are run. However, it is more likely that they do not consciously recognize, let alone resolve, the logical inconsistency, because they believe his physical disability also renders him incapable of reasoning as an equal and therefore they are not committed to reason as a process.

In this case, paradoxically, it is the business owners who appear to feel unsafe being challenged by a man with more education than most of them. They did not appear to trust or respect reasoning as a process for handling disagreement: their statements were coercive. He did feel safe at a personal level, having dealt with most of them before, but his trust in the procedure rapidly slipped. He expected to question them, while they did not expect to be questioned. He has little trust in social institutions: government, the education system, and society at large have had many opportunities to impress on him that he deserves nothing except what government and charities provide. The business owners appear to trust government, but only as long as it backs them up. The disagreement cannot be resolved by argument, not just because some participants seem to be committed to non-rational persuasion, but because there is clearly a lack of shared background and practices, enough to show that there is minimal interpersonal trust, no process-based trust, and little or no institutional trust here.

As noted earlier, there is no deep disagreement in principle about whether people with disabilities have a right to start a business in the community of their choice: the law says they do. But what the settled principle does not do is address the epistemic injustice (Fricker 2009). To the local business leaders, the consultant has no credibility. He was not silenced, but even when he countered their attacks, nothing he said was given any weight. Fricker’s (2009) hope that familiarity would help unfortunately had made little difference. The entrepreneur had followed this strategy: in his previous job, he had gone to considerable lengths to connect personally with the business owners and charities. Although he had had a number of successes in becoming recognized as a person worth listening to, in general their familiarity with him did not reduce their deep prejudice.

This returns us to the need to investigate social, rather than individual, approaches to getting everyone heard. I would characterize this scenario as one in which the business leaders’ minimal political trust (McMurphy’s term) or institutional trust (Zucker’s term) pulls them into a process that is designed to limit “opportunities for insight and resolution” in order to prevent any “advance [of] social justice” (Hundleby and Rooney 2006).

For this disagreement to move back into reach of rational resolution, it would have to address issues of trust in connection with power and control in achieving social change. The legal system may seem to grant equal rights, but is it in practice able to honour them? This is a case in which resolving a disagreement intellectually and even legally does not end our responsibility as reasoners. Here, we have a clear example of “indirect political” responsibility in “striving for individual freedom, non-violence, intellectual pluralism, and institutionalized safeguards for rights and means to obtain information and pass criticism” (Van Eemeren and Grootendorst 288). We cannot place the onus for opening minds on the person who has been disrespected: that person is precisely the one who has lost trust that others will exercise their power responsibly and is not trusted by the people who need to listen. We take responsibilities as allies for finding equalizing methods to remove rhetorical disadvantages. We share a responsibility to build trust, using rational persuasion, additional trusted participants as intervenors, and anything else in the repertoire of conflict management that can bring radically opposed parties to the same table. For example, one move that can rebuild trust is to find an intervenor with the power to make sure the disadvantaged person has the opening move in deciding what to do next; this is often done in mediation. Another option is a round-table discussion with a balance between members of the disadvantaged group and members of the privileged, with a skilled facilitator. (This entrepreneur had used a similar process successfully in earlier community discussions of the sex trade and of LGBTQ rights.)

6.4 Scenario 4

In cases such as the dispute between Schiavo’s parents, husband, and doctors about her treatment, institutional trust is often present even if interpersonal trust is not. Shiavo’s parents and husband disagreed strongly. However, the doctors are legally required to give weight to the relatives’ opinions. If doctors and patients’ families cannot reach consensus about treatment on their own, they are required to yield to the decision of a hospital ethics board, or sometimes a judge. When the parties trust the social institution and procedure, they can, as Adams (2005) suggests, reason as best they can as long as time permits. Even if they initially disagree, they might still surprise themselves by feeling heard and finding common ground. If they cannot reach consensus, they can (or may be required to) turn their debate over to a legitimate authority for resolution.

Adams can justifiably recommend that this deep disagreement in beliefs about life support should not be treated as a situation in which reasoning must cease. But nor is it required that reasoned discussion must continue if any participant refuses. Trust becomes an important deciding factor. Because the social context sets up a process with an authority that can make a decision when consensus cannot be reached, reasoned discussion can end if participants become angry or silent. As with many dispute-resolution processes, the process can legitimately include an option for the group to declare it cannot reach a decision and must “book out” of the dispute. (Mediators do this when they see no prospect for the parties reaching a settlement.) It is less crucial that the treatment plan be resolved by consensus than that all parties meet their legal responsibility to play their required part in the institution of ethical medical decision-making. The institutional process ensures there will be a decision within the time limit, however deeply the participants continue to disagree on principle.

However, if some participants do not trust the institutional process because they lack trust in the medical or legal systems, the disagreement continues to be deep, and connects to a second deep disagreement about whether the medical or legal systems are sufficiently fair for the society in which they are used. As in Scenario 3, this could throw us back to hoping that we can see a way to amend the process to create more trust.

6.5 Scenario 5

Dare positions the vaccination issue as one of deep disagreement: neither side is persuaded by the reasons of the other. He argues that reasoning continues, because the government continues to accept good scientific reasoning for vaccination and uses rational persuasion techniques on the undecided. His argument that reasoning continues is not compelling. It is not clear that government initiatives to maximize the number of vaccinators constitute “rational persuasion” of the type envisaged by Godden (2013), because the government targeted undecided parents, not the most principled objectors to vaccination. This side-steps the disagreement, rather than continuing to work on it.

Trust helps here to clarify that the New Zealand government’s actions are justified not by its methods but by its legitimacy. As the elected government, it is generally trusted to act responsibly on its citizens’ behalf. If government can demonstrate that reason shows the risk of stopping the vaccination program is greater than the risk of continuing, it cannot be said to be abusing its power. If it has these reasons, then the disagreement about the value of vaccination is not a deep disagreement: it has been settled. However, the people who oppose vaccination are now in a similar position to the business owners and charities in Scenario 3: they do not accept the result of good reasoning. They take the reasoning process to be flawed. If they do not trust science while the government does, once again trust becomes an entry point for intervention. If they do not trust their government to be acting in their best interest, they enter a different debate: whether the “third order” conditions in their society are sufficiently inclusive and respectful. Dare presumes they are, but this is not a decision that can be made unilaterally even by a legitimate government.

The five cases can be mapped on the trust-reason diagram as shown below (Fig. 3).

Fig. 3
figure 3

Sample situations mapped in two dimensions

Full size image

6.6 Scenario 1

The discussion between friends, which foundered because of trust problems, I have placed on the edge of the “intractable” disagreements, near the “deep” disagreements. The process might be changed to create a more exploratory, coalescent approach, but this might only confirm that the difference with respect to the merit of identity politics is a deep disagreement. Invoking trust as a dimension of this disagreement and as a measure of its success is that a discussion of trust helps the friends decide what their responsibility is. If their friendship could be jeopardized by this particular difference in beliefs, they may terminate the discussion. If they choose to continue the debate, their choice of an improved procedure will be selected not just for its logical accuracy and its methods for introducing and weighing evidence but also for its affective dimension, its ability to maintain interpersonal trust.

6.7 Scenario 2

The disagreement between colleagues is one in which any loss of interpersonal trust was temporary. Procedural trust, however, was not restored, for the reasons explained earlier. It has been positioned as probably a deep disagreement on the original topic, whether to avoid classroom examples that upset students, because of differences in core beliefs about how freedom of speech works. However, it remains midway on the trust axis because there is good personal trust between colleagues, and between students and their professors in this department. Responsibility to continue discussion depends on clarification of whether the issue is departmental policy or student comfort in the classroom, and the subsequent determination of who should be involved in the discussion.

6.8 Scenario 3

The business leaders, in trying to stop the consultant, indicate a lack of shared background beliefs and lack of shared procedure, as well as no significant trust of any type. I have placed this as a deep disagreement in spite of the question of equal rights having been resolved at a legal level. Because the trust level is so low, this instance of epistemic justice is deep at a procedural and institutional level – neither the internal procedures for reasoning nor the external requirements for equality are agreed on. As Govier (1997) contends, we must hope that we can find ways to contribute to the reduction of prejudice, without putting any more pressure on the recipient of that prejudice. There is clearly a shared social responsibility to look for new ways to make it possible to eliminate injustice.

6.9 Scenario 4

An attempt to reach consensus between family members and medical professionals would vary over a considerable range from accessible to deep, and from high to minimal trust, depending on the individuals and the medical situation involved. If any family member does not trust the social context – the legal requirements that constrain the discussion – then the disagreement becomes both deep and intractable, yet as Adams (2005) proposed, discussion should continue as long as possible. It is social trust that allows a way forward here. As noted earlier, as soon as it becomes clear that there is a lack of interpersonal trust or procedural comfort, it should be possible for the group to declare itself unable to reach consensus on the medical decision, and refer the case to the hospital ethics committee. However, if institutional trust is also lacking, the situation resembles Scenario 3. We want to avoid epistemic injustice, so the institutional procedure needs to be examined for its fairness and for ways to make people who might have to participate feel more trusting, and trusted, when they do so.

6.10 Scenario 5

Government continuing a program in the face of profound opposition, also has a wide range of cases, in some of which trust factors may play a significant role in opening up options to proceed. The individual differences of opinion on vaccination connect with differences of opinion on whether a government may ever compel its citizens to act against their beliefs. As with both previous scenarios, treating trust as a factor in the disagreement opens up the possibility of dealing with the trust issues at the same time as checking differences of procedure or belief, which is one way of taking responsibility for increasing social inclusiveness and equality.

7 Conclusion

In all the examples considered above, being able to reason is not just a matter of personal disposition or even of learned competence. Even for people who are open-minded by disposition, or highly skilled in argumentative disciplines such as philosophy, the first two examples indicate that a disagreement can become impossible to reconcile if one person involved does not feel safe enough to talk comfortably. Since deep disagreements in the classic sense may nevertheless have to continue instead of being dropped, as illustrated by the examples described by Adams (2005) and Dare (2016), trust of some type offers us a way forward, a way to begin to deal with “third order” issues of social inequality in disagreements. In our determination of whether a disagreement is “deep”, we can examine not only how it shares characteristics of disagreements that have no rational resolution resources but also how the actual participants’ responsibilities must be carried out. Can they step out and wait for “ideal peers” to determination what is possible? Should they continue as best they can, with the safety net of a procedure they trust to make a decision when they cannot? Must they suspend proceedings but shoulder their general social responsibility to make it possible for all voices to be heard and all evidence fairly considered?

I cannot conclude that trust, responsibility, and reason must be connected in the way I have connected them here, but I would propose that the affective, procedural, and social dimensions of disagreement cannot either be eliminated in advance or be postponed until after the clash of beliefs is categorized as a deep disagreement. “Ideal peers” are insufficient as a replacement for difficulties in real disagreements.

Even though it may be challenging to act, and not just hope to act, for social change, our responsibility to maximize the effective use of reason requires us to look at the social aspects of reasoning processes. If we can connect deep disagreements with trust, and with conflict management in general, we can be more confident we are in fact making our most responsible efforts to resolve a disagreement with reason.