1. Introduction
When we trust someone, we are vulnerable to them – this is quite an uncontroversial point. Now the interesting question is: what and how exactly are we vulnerable? Annette Baier first characterizes this vulnerability in terms of reactive feelings we naturally associate with violations of trust, saying that “trusting can be betrayed, or at least let down” (Reference Baier1986: 235).Footnote 1 This point is followed by Richard Holton, who doubles down and distinguishes between trust and mere reliance through a disposition to having those feelings. Holton writes, “When you trust someone to do something… you have a readiness to feel betrayal should it be disappointed, and gratitude should it be upheld (Reference Holton1994: 67).Footnote 2
This view, which has become widely accepted in the literature, does more than say that we are vulnerable to experiencing personal disappointment or betrayal when we trust someone. The reactive-feeling view holds that these emotional dispositions are constitutive of trust itself: to trust someone is, by definition, to be disposed to feel betrayal if they fail to follow through and potentially gratitude if they do. These emotional dispositions are what distinguish trust from mere reliance, and so they are built into its very structure through the broader commitment to a participant stance.Footnote 3
In this paper, I challenge the reactive-feelings view of trust. I focus particularly on its negative instances, that is, cases of trust violations. I argue that feelings of betrayal and personal disappointment in response to trust violations are neither necessary nor sufficient conditions for trust. Instead, I propose that trust is a fully doxastic mechanism – i.e., an epistemic attitude – directed at the norms constitutive of relationships. Therefore, violations of trust only directly call for an epistemic reassessment of the trustee’s trustworthiness, similar to other cases of belief disconfirmation. I then go on to explain the recurrence of reactive feelings with trust violations, arguing that they are better understood as responses to relationship impairments. That is, when someone violates our trust, this violation often just reveals that the person disregards something fundamental about our relationship.
My argument goes as follows: First, in Section 2, I explain what I find problematic and counterintuitive about the reactive-feelings view. In Section 3, I argue that reactive feelings are neither necessary nor sufficient for trust. To build my positive argument, in Section 4, I outline the core tenets of a relational account of trust, drawing on Israel (Reference Israel2025b). Sections 5 and 6 form the heart of the paper, elaborating on how trust functions as an epistemic mechanism and how reactive feelings arise from relationship impairments, thereby explaining their appropriateness without taking them to be intrinsic to trust.Footnote 4
2. Why trouble?
We can quite easily accept the weaker claim that reactive feelings simply often happen in the context of trusting relationships. However, the stronger claim – that trust embeds and is defined by these feelings, as Holton argues when attributing this to the Strawsonian “participant stance” – is an overkill.Footnote 5
First, trust is widely agreed to be a success-oriented or “optimistic” attitude (Baker Reference Baker1987; Jones Reference Jones1996; Reference Jones2019). When we trust someone, we believe the trustee will follow through, which allows us to set aside skepticism and the need to make contingency plans.Footnote 6 This point is emphasized in many recent discussions: Keren (Reference Keren2014) argues that trust involves responding to “preemptive reasons” – that is, second-order reasons that exclude or override certain other first-order reasons for action or belief; Marušić contends that “You trust someone if and only if, in light of her commitment as such, you believe that she will follow through” (Reference Marušić2017, 5); Nguyen (Reference Nguyen2022) regards trust as an “unquestioning attitude;” and Chae (Reference Chae2022), who shows how making contingency plans to avoid violations undermines trusting relationships.
Given this success-driven character of trust, a charitable reading of Holton would be that trust does not require an active psychological anticipation of failure. Rather, the “readiness to feel betrayal” should be understood as merely a necessary disposition – the trustor is disposed to feel betrayal should trust be violated but doesn’t actively expect or accept this possibility.Footnote 7
Nevertheless, even when understood this way, the discomfort remains. Defining trust by its potential failure still makes the possibility of failure constitutive of trust, rather than, say, the special forms of cooperation that it enables. This concern is primarily conceptual: it is strange to define the concept of trust by its negative potential, just as it would be strange to define what a living organism is by what makes it sick or die rather than by its vital functions. A definition should, in principle, regard how the phenomena is successfully instantiated, rather than how things would look like if it failed.Footnote 8
Furthermore, the mere addition of a disposition to disappointment or betrayal – which carries negative emotional weight – cannot be what differentiates trust and reliance if trust is to be appealing in any way. Trust and reliance are both attitudes that serve us in theoretical and practical reasoning, but trust seems to provide us with some unique advantage in our epistemic and practical projects. If, however, all that trust adds to reliance is making the stakes higher for the trustor, making them susceptible to feeling bad, then we’d take reliance over trust any day.
Holton gives us a view about trust’s value over reliance, but one that just doesn’t concern these projects in a substantial and direct way. He suggests that trust, as an attitude from a participant stance, is conducive to better relationships. He writes:
“Suppose we are rock climbing together. I have a choice between taking your hand, or taking the rope. I might think each equally reliable; but I can have a reason for taking your hand that I do not have for taking the rope. In taking your hand, I trust you; in so doing our relationship moves a little further forward.” (Reference Holton1994, 69)
Thus, for Holton, the advantage of trusting you over relying on a rope lies in what it would do to our relationship. Now, nurturing relationships is certainly important (this is especially true of my view, as I lay out in section 4), but can it be enough of a reason to decide to trust? In low-stakes scenarios, sure – I care about advancing our relationship “a little further forward.” But in rock climbing? Where a failure can cost my life? In such a high-stakes scenario, this doesn’t seem like a wise choice.
Deciding to trust always adds a contingency: in addition to capacity (the rope’s and your capacity to carry my weight), we suddenly also depend on the trustees’ will and even on their attention. You and the rope may be equally reliable in carrying my weight, but the rope can’t be distracted by a bird passing by or suddenly led by the evil intrusive thought that if it let me fall, it could marry my wife – but you certainly can!Footnote 9 Trust, in this way, is more fragile – it can fail in ways that reliance is less susceptible to. Given these additional vulnerabilities (and adding to it the extra emotional burden of feeling betrayed/disappointed), if all that trust adds to reliance as its contribution to the relationship, we should still prefer the latter, at least when it really matters.Footnote 10
A second concern is that reactive-feeling views are in tension with fully doxastic accounts of trust, which are often referred to as cognitive views (McMyler Reference McMyler2011).Footnote 11 Those have become prevalent in the literature for several reasons. First, they provide a unified explanatory framework that connects trust to our broader epistemic practices – explaining how trust functions in testimony and knowledge transmission, as well as in practical reasoning and coordination of action (Hardin Reference Hardin2002; Hieronymi Reference Hieronymi2008; McMyler Reference McMyler2011; Hawley Reference Hawley2014).Footnote 12 Second, doxastic accounts preserve trust’s rational character by making it responsive to evidence and reasons, rather than treating it as a mere disposition or emotional stance. This responsiveness allows us to distinguish appropriate from inappropriate instances of trust based on evidence concerning the trustee’s trustworthiness. Finally, by grounding trust in beliefs about another agent’s trustworthiness, these accounts can talk about trust in terms of justification and account for when and how it can be rationally revised or withdrawn.
The tension between reactive feelings and doxastic accounts comes from the following: if trust consists of a belief regarding the truth of a trustee’s assertion or commitment to act in a certain way, then a violation of trust would lead us only to engage in epistemic reassessment – determining whether the belief remains justified and the trustee is still to be held trustworthy (McMyler Reference McMyler2011, 131). If that is the case, it is not clear how we can accommodate this purely epistemic reassessment with a view that takes emotional responses to be a necessary component of trust violations.
To understand what this means, we may consider the distinct roles of beliefs and emotional responses: beliefs are cognitive states that can be evaluated for truth and evidence, and so when new information comes in, we usually update our beliefs. Emotional responses, on the other hand, are usually conceived as reactive states that may or may not track reasons. Suppose, for example, that you’re a scientist and you have high confidence in your hypothesis. If it turns out the data doesn’t support your hypothesis, it’s surely reasonable for you to feel disappointed and even angry. Again, emotions are connected to our beliefs in various ways, and that is particularly true in response to their confirmation or disconfirmation (Reisenzein Reference Reisenzein2009a; Reference Reisenzein2009b; Miceli and Castelfranchi Reference Miceli and Castelfranchi2015). Nevertheless, this connection is not intrinsic nor necessary to the belief – the epistemic process directly concerns only the evidence and the appropriate assessment and process of it.
Thus, when the data proves your hypothesis wrong, your direct cognitive response is to revise your belief about the hypothesis’ truth based on evidential grounds. Your emotional response (disappointment/anger) is secondary, not fully governed by standards of evidence or truth-aptness, and doesn’t take a necessary role in your epistemic project.Footnote 13 In other words, while it is reasonable for you to feel disappointed or angry when your hypothesis has been proved wrong, you could as well not feel that way for what concerns your epistemic aims as a scientist. The puzzle for doxastic accounts of trust is that according to reactive-feeling views, trust violations hold a necessary relation to emotional responses, exceeding mere epistemic reassessment. However, if trust is purely belief-based, we would expect violations to lead to epistemic reassessment of the kind we see in the scientist case, and if they elicit any emotional reactions, those should take a similar form to the scientist’s as well – being secondary, not fully governed by standards of evidence or truth-aptness, and contingent.
Now, this is a problem that specifically challenges fully doxastic accounts. It doesn’t, however, undermine impure doxastic ones (which take the relevant belief to be necessary but not sufficient for trust), if what they add to belief is something that can accommodate affective responses, such as goodwill (Baier Reference Baier1986), optimism (Jones Reference Jones1996), or a participant stance (Holton Reference Holton1994; Marušić Reference Marušić2017). Nevertheless, if we can explain these emotional dimensions without positing additional elements to trust beyond belief, we would achieve a more elegant and parsimonious account. For this reason, in this paper, I engage in resolving this tension while maintaining a fully doxastic account of trust, demonstrating how the rich emotional responses to trust violations can be explained within a purely doxastic framework, alluding instead to what happens in the context of trust: the relationship.
3. Reasons to reject the reactive-feelings view
To keep you on track, what I argue for in this paper is that while there is quite a consensus that reactive feelings are an inherent characteristic of trust, this is only because we often conflate the characteristics of trust with those of the relationships where it takes place. Reactive feelings, I contend, belong to the latter.
Besides the challenges laid out above, in this section, I start the negative part of my argument (why we should reject the reactive-feelings views) by arguing that reactive feelings do not hold a necessary nor a sufficient relation to trust.Footnote 14 In addition, I contend that to distinguish warranted instances of feelings of betrayal vs. personal disappointment, we are prompted to look beyond the violation itself and toward the relationship where it takes place.Footnote 15
To show how trust does not necessarily involve a potential for betrayal, let’s think first of non-personal forms of trust, such as trust in institutions, where the disconnection between these becomes quite salient. Imagine you trust your bank to safely hold your deposits and process transactions. When the bank’s system experiences a server failure and causes your mortgage payment to bounce, you may reasonably reassess the bank’s operational competence and might switch to another institution. As Alfano and Huijts argue, institutional trustworthiness requires competence with respect to the institution’s domain and the ability to reliably signal to others their competency (forthcoming, 8), and the bank has failed you in both aspects. This failure serves as evidence against the bank’s trustworthiness, possibly warranting an attitude of distrust toward the bank (forthcoming, 11). Nevertheless, feeling betrayed by this institutional failure seems misplaced. This may be somewhat due to the non-personal nature of the interaction, which makes claiming you feel betrayed and personally disappointed seem like a misuse of language and poor emotional management.Footnote 16
The same disconnection between trust and reactive feelings can also be found in instances of personal trust. Therapeutic trust is a good case for us to look at for this purpose. It represents scenarios where trust doesn’t reflect a belief that the trustee will carry through their commitment and perform the expected action successfully, but instead it aims at growing trustworthiness in them (Nickel Reference Nickel2007, 318; Carter Reference Carter2024; Horsburgh Reference Horsburgh1960; Jones Reference Jones, DesAutels and Walker2004). McGeer illustrates this form of trust with the example of “parents deciding to trust their teenagers with the house or family car, believing that their offspring may well abuse their trust, but hoping by such trust to elicit, in the fullness of time, more responsible and responsive trustworthy behaviour” (Reference McGeer2008, 241).
Arguably, trust can’t be betrayed in such cases as the trustee is expected, at least up until a certain point, to fail.Footnote 17 For this reason, some scholars like Hieronymi (Reference Hieronymi2008) contend that therapeutic trust does not qualify as true or full-fledged trust, suggesting that the trustor merely acts “as if” they trust. I think, however, there is a plausible story to tell about the beliefs implied in such cases. It may not be a belief that the trustee will follow through with the particular action at hand, but instead, as Keren argues, that “the trusted person has a degree of trustworthiness” and a belief of “that our trusting her will help her become more trustworthy” (2020, 116–17).Footnote 18
As for whether reactive feelings are a sufficient condition for trust, I argue that, similar to how therapeutic trust represents a form of trust devoid of reactive feelings such as feelings of betrayal or personal disappointment, we may also find cases where those feelings are warranted while trust is not to be found in the relationship. For example, consider a married couple who, after decades together, have settled into cold coexistence. Love has long faded, trust even before that, but they stay together out of inertia and fear of starting over. Even in such an unfortunate scenario, it would still be reasonable for the wife, upon discovering that the husband has a girlfriend, to feel deeply betrayed, even though she hadn’t trusted him for years and, as a matter of fact, held no expectations of his fidelity. Her feeling of betrayal is grounded in the fact that she was factually betrayed. Hawley makes this point even stronger with an example of betrayal in cases of distrust. She argues that Jesus knew that Judas would betray him, which makes trust impossible, and yet he was betrayed by Judas and thus was justified in feeling betrayed (Reference Hawley2014: 13).Footnote 19
An additional worry that may be helpful in indicating the kind of relation between trust and reactive feelings is that we can’t properly distinguish between instances where disappointment is warranted rather than feelings of betrayal (and vice versa) without appealing outside the violation of trust.Footnote 20 Baier contends that all trust must be susceptible to betrayal and holds mere disappointment to instances of mere reliance on someone (Reference Baier1986: 235). This description fails to grasp many mild cases where the trustee’s failure merely justifies disappointment, perhaps because the matter in question was very minor or because we realize our expectations of them were unreasonable. Attempting to distinguish better between violations of trust that warrant disappointment and betrayal, Edward Hinchman expresses a similar concern for distinguishing when each attitude is appropriate. He argues that disappointment occurs when the trustee simply fails to do what they were trusted to do. Hinchman accounts for betrayal, contending that it operates “at a deeper level” when the trustee fails to properly engage with and respond to the trusting person’s needs in the way demanded by their shared understanding of the trusting relationship (Reference Hinchman and Simon2020, 58–59). This direction sounds promising, as betrayal is accounted for in ways that go beyond the particular violation, but it is not clear what makes a particular failure of trust operate at a “deeper level” that reaches to one’s understanding of the trusting relationship.
4. How should we think about trust?
If the objections above hold, then we would benefit from a theory of trust that explains the occurrence of reactive feelings but does not hold them as necessary nor sufficient for trust (in response to the first and second concerns raised) and explains why some breaches of trust warrant feelings of betrayal, while others warrant mere disappointment, thus providing an account of appropriate reactive feelings (in response to the third concern). As indicated by Hinchman’s discussion above, the appropriateness of reactive feelings seems to lie in the context where both trust and its reactive feelings take place: relationships.Footnote 21 Accordingly, we will look now at a relational account of trust that may be helpful with the challenges listed above,Footnote 22 an account that considers trust not as an isolated attitude but locates it in the interpersonal or social contexts where agents connect and cooperate with each other.
Trust requires the trustor to be in some kind of relationship with the trustee and implies a belief that the trustee will follow through with what the relationship entails (Israel Reference Israel2025, 13). By “relationships,” I don’t mean only the thick connections we have with family, friends, and loved ones. We normatively depend on and cooperate with other agents in ways that go far beyond our most intimate domains and also include many thin connections, like the ones we hold with colleagues, teammates, fellow citizens, and even fellow human beings (members of a shared moral community) – each carrying its own set of special duties and responsibilities (Scheffler Reference Scheffler and Scheffler2002, 97).Footnote 23
Relationships, insofar as they can accommodate trust, have a normative component. That is, norms of conduct apply – ways in which we are committed to acting with each other – as part of the relationship. This view of relationships is supported by Raz, who contends those are “constituted by norms which determine what conduct is appropriate between people in the relationships, between, for example, parents and children, as well as towards other people’s children, etc.” (Raz Reference Raz and Raz2009, 316). These norms or rules aren’t merely part of relationships but are constitutive of them, and “our knowledge of these norms, that is of the nature of the relationship, guides our behaviour within it” (2009, 317).
At a foundational level, the norms of a relationship establish its character – whether friendship, professional association, or kinship. For instance, what makes a relationship a friendship rather than a mere acquaintance is not just the frequency of interaction but the normative expectations that define friendship itself. More importantly, to our subject matter, relationship norms provide practical guides, establishing much of what participants owe to one another. These norms explain why we can reasonably expect certain forms of care from a doctor that we wouldn’t from a neighbor, why we may form scientific knowledge from what a scientist says but not from what a social media influencer preaches, or why we might expect emotional support from a friend but not from a store clerk, constituting what Scanlon refers to as the “ground relationship” (Reference Scanlon2008, 155) that establishes the expectations against which many primary deviations are measured and constitutes the vulnerability agents are in when they relate to one another.
This level is shared by both personal and non-personal forms of trust. In institutional contexts, such as the one I raised in Section 3, these constitutive norms often take a more codified or explicit form, such as in professional codes and bylaws, but may still include the implicit social expectations embedded in a sense of “the way things are done” (Delacroix 2017). When we trust a bank, we don’t merely predict that their servers will function, but we hold the institution to normative standards that constitute the “client-institution” relationship.Footnote 24
Beyond what the ground relationship tells us, some of the norms that concern trust are shaped by the agents, personalizing the ground relationship and making it unique. These norms reflect what Raz describes as the way relationships are not only “socially structured” but also “individually explored and established” (Raz Reference Raz and Raz2009, 317). For instance, what counts as appropriate communication in one friendship may differ substantially from another, based on the particular history and the conduct established between particular friends. As noted by Israel, “through repeated practice, trustors and trustees develop an understanding of how to apply rules in various contexts” (Israel Reference Israel2025, 20).
By shifting the focus from actions to relationship norms, this view offers a unified framework for understanding how trust operates across different kinds of relationships. It does so by identifying a common underlying structure: relationships that are constituted by norms. In all instances of trust, the trustor holds a belief that the trustee will follow through on the norms that define their relationship. This allows us to treat trust as a single epistemic – or doxastic – mechanism, despite the variability of its content across contexts (different kinds of relationships, which we will explore further in the following section). On standard doxastic accounts, trust implies a belief that the trustee is trustworthy (McLeod Reference McLeod, Edward and N.2023). Under the relational version I adopt here, trustworthiness consists in the trustee’s norm-guided participation in the relationship. That is, when we interact within a trusting relationship, we treat the trustee’s conduct as evidence for or against their commitment to the norms in force. This unified mechanism allows us to capture both the shared epistemic structure of trust and the diversity of expectations that arise across different relational contexts.
5. Trust as a doxastic mechanism
With this account of trust in hand, we may now turn to explore what happens when trust is violated. A key point of my argument lies in the distinction between the doxastic mechanism of trust and the plural character of the relationships in which it operates. Let us unpack the former first.
When trust is violated, the violation may stem from a localized mistake or a misunderstanding of expectations on the part of the trustee – one that doesn’t necessarily reflect on their broader trustworthiness and thus doesn’t warrant reassessing our belief.Footnote 25 Suppose, for example, you trusted a friend to water your plants while you were away, and upon returning, you find them all dead. You’d likely be upset, but after learning that she overwatered them – perhaps due to having experience only with other kinds of plants or a misreading of your message saying to water them “around 3 times,” which she interpreted as per day rather than per week. In such a scenario, you would reasonably conclude that it was an honest mistake, and the failure does not seem to reflect on her general trustworthiness.
Note here the difference between the anger you feel about the dead plants and the potential to feel betrayed depending on the reasons for that outcome: your anger seems to be grounded in the facts (that your plants died or that your friend’s actions are the causal reason for that), and you may as well be justifiably angry even after learning the details about what caused your friend to overwater your plants. The feelings of betrayal or personal disappointment, however, seem to be grounded elsewhere, in what the friend’s action says about their care and level of commitment to you.
Similarly, a trustee might be genuinely committed to the relationship’s norms while still developing their norm/rule-following capacities, as we see in cases of therapeutic trust. Think again of parents trusting their teenage son with the house for the weekend. Upon returning, they learn that the son had a girl over despite having agreed not to do so, but the house is clean, nothing is broken, and he even admitted it upfront. While this goes against what was entrusted, it doesn’t suggest a lack of regard for the relationship or its norms – rather, it reflects a judgment lapse within the bounds of what the parents could reasonably expect, given the son’s degree of trustworthiness (which again, they are working on).Footnote 26 Here, the violation falls within the margin of error anticipated by the trustors (the parents). This is not to say that epistemic reassessment does not occur; rather, because the trust is directed at the trustee’s potential to develop trustworthiness, the bar for revising the belief downward is higher, and the parents expect local failures as part of the learning curve. However, this belief is not immune to updates. For instance, a persistent pattern of violations would eventually wear out the trust, providing sufficient evidence that the trustee is not participating in the developmental process at all.Footnote 27
However, some violations provide direct evidence against a trustee’s trustworthiness, indicating their inability to properly engage in relationships of the relevant kind or simply their disregard for the particular relationship at hand. Consider a romantic partner’s deliberate decision to have an affair. This decision and the acts that follow it represent a clear violation of a constitutive norm of monogamous relationships and reveal a fundamental disregard for the relationship itself. Alternatively, suppose that a political leader (fill in your own), during a public crisis, prioritizes their popularity and political allegiances over their citizens’ security and well-being. These intentional violations call for trustors (the betrayed partner or the citizens) to reassess their belief in the trustworthiness of the trustees (the unfaithful partner or the political leader).Footnote 28
Insofar as we discuss trust as an epistemic mechanism of cooperation, this is all that needs to be said: when violations occur, they may serve as evidence regarding the trustee’s trustworthiness and may or may not impact the trustor’s degree of belief. By mechanism of cooperation, I mean that trust functions as a belief-based process that enables cooperation in relationships, promoting our epistemic and practical projects. Whether or not the trust belief is to be reassessed depends largely on the reasons for the apparent violation. Namely, when the reasons for the violation are local (i.e., concerning a failure in execution or understanding in a particular action), we do not seem prompt to reassess trustworthiness. However, when the violation concerns the relationship norms, that may deem the trustee less trustworthy in the relationship.Footnote 29 In the event of the latter, the violation may lead us to engage in epistemic reassessment. The reassessment of the trustee’s trustworthiness in light of a violation concerns solely the effectiveness of cooperation, and thus, it is sensitive to evidence, as to be expected from a doxastic mechanism.
6. Relocating reactive feelings
So far, I have demonstrated how trust, as a relational doxastic mechanism, only goes so far as allowing epistemic reassessment when violated.Footnote 30 That is, it does not necessarily move us to change the way we feel or regard the trustee who violated our trust. Instead, it merely creates the grounds for such changes if we come to believe less of them in light of their behavior. Thus, reactive feelings like disappointment or betrayal cannot be inherent to trust or conceptualized as a distinctive characteristic of the attitude of trust.
This leaves us with some pressing questions regarding the occurrence of reactive feelings. First, where do these reactive feelings belong? Second, how can we account for the widely endorsed (and largely correct) observation that such feelings often happen in light of trust violations? Finally, what makes these feelings warranted if not the violation of trust itself? In this final section of the paper, I suggest that these reactive feelings are tied to the relationship. Thus, a particular failure in trust may lead to impairments or changes in the broader relationship.
As pointed out in the preceding section, there is a substantial difference between violations that are unintended or expected and violations we take to be done in disregard for the norms that constitute the relationship. For the former, we do not seem to be immediately prompted to update our belief on the trustee’s trustworthiness, while for the latter, we do take the violation as the right kind of evidence for rethinking our trust in them. This difference, however, lies not in trust, but in the relationship itself.
To understand this difference more clearly, we should distinguish between two distinct (albeit interrelated) evaluative processes. While in the phenomenology of our everyday experience these assessments often occur simultaneously and rely on the same content, they differ in their aim and orientation. When a trust violation occurs:
-
1. We engage in an epistemic assessment of the violation and its circumstances, forming a judgment about the nature of the violation (e.g., if it was an honest mistake, due to local incompetence, intentional, etc.). This judgment is primarily forward-looking: it informs our belief about the trustee’s trustworthiness and concerns the viability of future cooperation.Footnote 31
-
2. Separately, we engage in a backward-looking relationship assessment. Here, we evaluate whether the violation reveals incompetence or disregard toward the relationship and its constitutive norms. Here, we’re not assessing whether they’re likely to follow through in the future, but whether they value and are truly committed to what our relationship entails.Footnote 32
Trust as a doxastic mechanism is concerned with the epistemic process that guides our belief in the trustee’s trustworthiness. Reactive feelings, however, arise from the relationship assessment process in response to judgments about the trustee’s regard for the relationship. To further illustrate these processes, let us consider again the plant case. In the epistemic assessment process, when your friend fails to keep your plants alive, you may judge that she reasonably misunderstood your instructions rather than being careless or indifferent. This judgment doesn’t (at least significantly) have to undermine your belief in her trustworthiness.
Simultaneously, in the relationship assessment process, you evaluate what this violation reveals about your friend’s attitude toward your friendship. You may consider factors like whether she made a genuine and sufficient effort to care for the plants, if your expectations of her were clearly laid out, whether she was reasonably expected to know better, and even if she apologized sincerely and tried to fix things in some way. After considering such factors, you might judge that she values our friendship and its norms, obviating the need to reassess the relationship. However, if through this process, you found that she never bothered to check on the plants or just overwatered to avoid coming over to your house multiple times, then you may conclude that her violation reveals a lack of regard for your friendship and its norms, constituting an impairment in the relationship.
In the latter alternative scenario, reactive feelings come in, responding to relationship impairments. Relationships, as noted, are plural in character. By plurality I mean, first, that relationships may take a variety of forms. We have multiple kinds of relationships, with different people, with different obligations and different levels of commitment.Footnote 33 However, it also means that there is more to relationships than just practical expectations. For instance, there are feelings we may feel when someone doesn’t regard us as we would rightfully expect or wish them to. It follows then that the reactive feelings that are often associated with trust, such as disappointment and betrayal, are not truly characteristic of trust, and trust does not entail “a readiness to feel betrayal” (Holton Reference Holton1994, 67). Violations of trust warrant only judgments and reassessments about the trustee’s trustworthiness and may lead to distrusting the trustee depending on the recurrence or significance of the break and the aims of trust.
While violations of trust may only directly ground our assessment of a trustee’s trustworthiness, guiding us on whether it is appropriate/justified to trust them, I argue that trust-related reactive feelings are responses to relationship impairments. Violations of trust may indeed reveal something about one’s attitudes toward the relationship and even directly impair the relationship, as they consist of violations of the norms constitutive of the relationship. However, reactive feelings only become appropriate to the extent the relationship itself was impaired.
Accordingly, one may experience similar reactive feelings when trust is not present. For instance, consider again the case of the trustless married couple from section 3 – trust does not define the relationship between the couple, but the fact they are committed by their marriage vows to sexual exclusivity warrants feelings of betrayal when an affair is revealed. Similarly, to draw again on the example of political trust, one can appropriately feel disappointed and even betrayed by its government even if they did not trust their representatives to begin with.
This takes us back to the distinction between the mechanism of trust and the plurality found in our relationships that I opened this section with. Under the account I advance here, trust is a fully doxastic mechanism that targets effective cooperation within relationships. As a mere mechanism, it requires content within which it operates, and that content is found in the relationships where trust takes place. Whenever something goes wrong with trust, there are ways in which the mechanism needs to address what might be an epistemic failure, that is, reassessing the nature of the violation and whether the trustee is still to be held trustworthy.
Now, if we look at it just as to how trust works, it seems to be indistinguishable from mere reliance. However, we should not forget that its content – the relationship norms – is inseparable from it. Trust is a form of reliance, in the sense that it is a cognitive attitude that guides one’s reasoning, such that “relying on p involves a disposition to, among other things, deliberate on the basis of p, plan on the basis of p, act on the basis of p, and draw conclusions from p” (Alonso Reference Alonso2014, 166).Footnote 34 . However, one cannot trust without a relationship with norms that guide the parties’ conduct toward one another, providing it with the particular kind of indeterminacy and agential stance that we value and nurture, allowing for trustee discretion and accountability.
7. Conclusion
In this paper, I have challenged the prevailing view that reactive feelings, such as personal disappointment and betrayal, are inherent to trust, showing how it faces serious difficulties, primarily in accounting for cases where trust exists without those feelings and vice versa. As an alternative, I have advanced a relational account of trust where trust is directed toward norms constitutive of our relationship. This view allows us to keep trust as a fully doxastic mechanism of cooperation while avoiding the tensions that typically arise between doxastic accounts and the emotional responses associated with trust violations.
The act of “Breaking Trust,” as stated in this paper’s title, doesn’t only allude to my focus on trust violations but also to the move of setting apart trust as a mechanism from its content in relationships. – Under this account, violations of trust directly warrant only epistemic reassessments of trustworthiness – a purely cognitive adjustment that aligns with trust’s doxastic nature; and reactive feelings are to be relocated to occurrences of relationship impairment. With this separation, we can maintain a parsimonious account of trust (avoiding hybrid or pluralistic notions) while still accounting for the emotional responses associated with it.
Note that I don’t deny Holton’s basic insight. It would be foolish to do so given the recurrence of betrayal and personal disappointment along with trust violations. Holton is right that reactive feelings occur in the same environment of trust violations, and that environment (the relationship) requires holding a participant stance of the kind that Strawson advances, granting reactive feelings. It is just the particular necessary tie between trust and reactive feelings that needs to be undone, as those represent two distinct features of our relationships.
The breakdown of what happens in scenarios of trust violation explains not only the co-occurrence with reactive feelings by pointing at their shared object: the relationship norms. It may also shed light on what warrants and makes appropriate one kind of reactive feeling over another. Speaking to our focus on violations, it may better explain why some breaches of trust warrant feelings of betrayal, while others warrant mere disappointment, alluding to what Hinchman intended when making the distinction in terms of deeper and superficial levels where the violation occurs.
When someone violates trust, they violate norms constitutive of the relationship – norms that they were trusted to follow. What matters regarding the appropriateness of reactive feelings is not the violation of trust itself but what this reveals about the trustee’s attitudes toward the relationship and how the violation affects the ways in which the parties interact. We may distinguish between betrayal and disappointment based on the nature of the relationship and the impact of the particular violation to its integrity.
Regarding the first, some relationships, by their very nature, are thin and can’t accommodate feelings of betrayal because they lack the requisite normative structure or depth. For instance, if I trust my dance partner not to make inappropriate advances during a dance class, and they violate this trust by making unwanted sexual advances, blame and deep disappointment would be appropriate responses, while feelings of betrayal would generally not be warranted.Footnote 35 This is not because the violation wasn’t severe (it certainly was), but because the relationship itself typically lacks the normative depth that would make betrayal an appropriate response. In contrast, I trust my therapist with intimate details about my personal life and mental health. If they violate this trust by sharing these details with others, feelings of betrayal would be appropriate. The relationship between therapist and client, while professional, carries significant normative weight regarding confidentiality and care in ways that some violations may occur, as suggested by Hinchman, “at a deeper level.”Footnote 36
Regarding the significance of impairment – even within relationships that can accommodate betrayal, not all violations warrant such an intense reactive attitude. What matters is the degree to which the violation impairs the relationship by revealing attitudes fundamentally incompatible with it. A spouse’s infidelity, for example, typically warrants feelings of betrayal because it reveals attitudes fundamentally incompatible with the constitutive norms of monogamous marriage. Such violations of sexual and/or emotional exclusivity strike at the heart of what constitutes the relationship itself. On the other hand, a spouse bailing on a lunch date, while disappointing, generally wouldn’t warrant feelings of betrayal. This violation, while certainly a failure to meet relationship expectations, doesn’t reveal attitudes fundamentally incompatible with the relationship.
With this new setup, we may explain why therapeutic trust, as discussed in Section 2, does not immediately generate feelings of betrayal when the trustee fails. In such scenarios, the trustor suspends or calibrates differently the relation between the local violation and the relationship. The parent who trusts their teenager with the family car, expecting them to possibly fail, isn’t betrayed when failure occurs because the relationship impairment itself was anticipated as part of the developmental process. They may, however, feel disappointment, or even betrayal, if the violation regards the process of developing trustworthy conduct, or otherwise, if the teenager reveals a more profound disregard for their parents along with the particular violation.
We may also explain now cases where betrayal occurs without prior trust, such as the married couple example from Section 2. What matters for betrayal is not whether trust was broken but whether someone’s conduct reveals attitudes that disregard the relationship. In the case of the married couple who have settled into cold coexistence, the relationship, however damaged, still exists, and its norms are still binding. The violation of this norm significantly impairs whatever relationship remains, warranting the wife feeling betrayed.
Thus, understanding reactive feelings as responses to relationship impairments rather than to trust violations per se also helps explain their appropriateness. If betrayal were inherent to trust, then feeling betrayed would be rational whenever trust is violated in a significant way. But this doesn’t match our intuitions that mere nonconformity with expected actions is insufficient. We care about the reasons for a violation, and that is because we understand that not every failure to act as expected reflects a violation of the relevant norms, and also because some violations of the norms may be justified given the circumstances, in the sense that they do not warrant a reassessment of the relationship itself. Personal disappointment and feelings of betrayal are thus appropriate when they accurately track how violations reveal relationship-impairing attitudes.
Finally, the account in this paper may offer some interesting implications to be explored in the future, such as how to address relationships after a violation of trust. If reactive feelings target relationship impairments rather than trust itself, then repair efforts should focus on addressing the attitudes that caused the impairment and/or revisiting the norms that characterize the relationship, rather than simply rebuilding trust over flawed foundations. Namely, trust violations may tell us something about the trustee’s trustworthiness, but failures of trustworthiness in one domain or relationship do not necessarily entail failures in another. Sometimes, if beyond the particular relationship we find value in maintaining a relationship with the other person, changing or establishing the relationship is what’s really at stake, and so trust may be restored within a different relationship without being the primary target of repair.
Acknowledgments
For helpful comments on earlier versions, I thank Lee-Ann Chae, Eugene Chislenko, Sanford Goldberg, Miriam Solomon, and the anonymous reviewers for Episteme. I am also grateful to participants at the “From Bitterness to Disappointment and Back” workshop at the University of Vienna and the “Liberalism Rekindled” research project at the Hebrew University for valuable discussion. This research was supported by the Institute for Humane Studies under grant no. IHS018810.
Open access
