Trusting groups

ABSTRACT

Katherine Hawley was skeptical about group trust. Her main reason for this skepticism was that the distinction between trust and reliance, central to many theories of interpersonal trust, does not apply to trust in groups. Hawley’s skeptical arguments successfully shift the burden of proof to those who wish to continue with a concept of group trust. Nonetheless, I argue that a commitments account of the trust/reliance distinction can shoulder that burden. According to that commitments account, trust is a distinctive kind of reliance grounded in a positive appraisal of features of the trustee’s practical rationality, foremost their commitments and their capacity to act on those commitments. This is one way we can make sense of the difference between trusting and relying on individual people. I argue it is also a way we can make sense of the difference between trusting and relying on groups.

KEYWORDS:

• Trust
• reliance
• groups

1. Introduction

Katherine Hawley was skeptical about group trust (Hawley, 2017). Her main reason for this skepticism was that the distinction between trust and reliance, central to many theories of interpersonal trust, cannot be parsed when considering trust in groups. Hawley’s denial of group trust has received much less attention than her other work on trust, but her group-trust skepticism could have significant consequences.¹ A significant body of research on trust in the social sciences focuses on trust in complex collectives, such as institutions and organizations, and for many years this work has aligned with growing concern in public discourse about low levels of trust in, for instance, governments, banks, police, and the media.² But the proper remedies for our crises of public trust could look very different if Hawley’s skepticism is well-grounded. Perhaps the real crisis is a failure of trustworthiness among the individuals running these institutions, and not the institutions themselves. Or perhaps what we really want from these institutions is not trustworthiness at all, but rather reliability. Either way, how we understand these social problems and how we respond to them might need to be significantly altered if, as Hawley recommends, we abandon the concept of group trust.

I propose to take Hawley’s group-trust skepticism seriously and engage in some detail with the arguments she used to support it. As with all of her work on trust, Hawley’s analysis of group trust is thoughtful, insightful, and stimulating. Hawley’s skeptical arguments do indeed shift the burden of proof to those who wish to continue with a concept of group trust. Nonetheless, I believe with the right account of the group trust/reliance distinction, this burden can be shouldered. Specifically: I argue that a commitments account of the trust/reliance distinction can extend to groups while avoiding the problems that Hawley identified.

My response to the details of Hawley’s group-trust skepticism will occupy the first half of the paper (sections 2 and 3). The second half takes the lessons learned from both the successes and shortcomings of Hawley’s arguments as a starting point for my own account of a group-trust/reliance distinction. I explain how we can make such a distinction regarding interpersonal trust in section 4, and in section 5 I argue that we can translate this distinction to groups. Finally, in section 6 I argue that this distinction is not only intelligible, but valuable, in virtue of the fact that it tracks the distinctive value of trusting relationships between individuals and groups, and thereby vindicates public and academic concern with low levels of trust in political and social institutions.

A note on the term “group” before proceeding: Hawley uses the term to refer to a very wide variety of collective entities, from crowds gathered in one place (Hawley, 2017, p. 237), through more formalized collections of individuals sharing some activity (e.g., a group of students; op cit. p. 235), to more structured collectives such as organizations, corporations, and institutions (op cit. p. 241). In the sections below I follow Hawley’s usage of “group” as a general term for a variety of collective entities, though I focus primarily on internally complex groups that some might prefer to refer to as “organizations”. The benefit of focussing in this way is that my account explains the nature and value of trust specifically in many of those groups that are most relevant to our concerns about social crises of trust: governments, banks, media organizations etc. The drawback is that differences between these institutions and more loosely structured groups, such as a crowd gathered at a country fair (op cit. 237), could make it difficult to extend my account to all varieties of groups. Nonetheless, as I argue in more detail in section 6, there is value in rescuing a concept of group-trust even with this restricted scope.

2. Hawley’s epistemic group-trust skepticism

I will begin with Hawley’s skepticism about epistemic trust in groups. Trust has played a crucial role in a number of theories of testimonial knowledge.³ Epistemologies of testimony include two broad categories of theories: reductivist theories, which roughly speaking treat testimony as just another form of epistemic evidence; and non-reductivist theories, which take testimony to be in some sense sui generis. One variety of non-reductivist theories are assurance views, which maintain that when A tells B that P, A assures B of the truth of P and, given the right kind of interpersonal relationship between A and B, B may take this assurance as a reason to believe that P (Ross, 1986). Among these assurance views are some that are specifically trust-based, which maintain that for B to rationally accept the word of A their relationship must involve trust (Faulkner, 2007; Hinchman, 2005; Moran, 2006).

Hawley grants the plausibility of trust-based assurance views for interpersonal testimonial knowledge, but argues that they do not extend to group testimony, and that this gives us strong reason to doubt that we need a concept of group-trust to understand our epistemic interactions with groups. She considers a range of assurance views in application to groups, but to illustrate the shortcomings of her skepticism I will concentrate exclusively on her analysis of Moran’s account.

Moran (2006) observes that the epistemic status of testimony is distinctively threatened by the fact that testimony comprises words freely chosen by the speaker. Being told that P by a person is prima facie significantly different to a person presenting us with evidence for P, or a person showing us that P involuntarily; consider the difference between, respectively, court-witness testimony, a lawyer presenting CCTV footage, or a jury forming beliefs about a defendant from their body language. Given the agency a speaker exercises over their testimony, if we treat testimony as merely another form of evidence of the speaker’s beliefs we cannot but conclude that testimony is a bad form of evidence, vulnerable to speaker manipulation, and always second-best to better evidence. For Moran, assurance views are well equipped to recognize the possibility of rationally believing what we are told, because they can acknowledge that the act of telling a person involves the speaker standing behind their words, inviting the listener not just to trust what has been said but to trust the speaker themselves. If this trust is well earned, testimony provides a reason to believe not provided by other forms of evidence. When A tells B that P, A takes responsibility for the truth of P, and if B trusts A, then B has reason to take their word for it (Moran, 2006).

Hawley gives two reasons to think that Moran’s account is not so appealing when we switch to groups (Hawley, 2017, p. 242). First, Hawley argues that since it is far from uncontroversial that organizations have beliefs, a reductivist approach to group testimony is not committed to treating testimony as evidence of the beliefs of the group. Second, she maintains that groups do not freely choose the words of their testimony in the way that individuals do. Rather, according to Hawley, statements issued by organizations are produced “via the functioning of internal mechanisms” (ibid). And if these statements are not freely chosen words, Moran’s way of distinguishing testimony fails to apply in the group case.

But neither of these reasons give us good grounds for group-trust skepticism. Consider first Hawley’s observation that reductivist theories need not take group testimony to be evidence of group belief. Even if we grant this, it sidesteps Moran’s main concern: that when testimony is understood as evidence like any other, regardless of what it is evidence for, its vulnerability to speaker manipulation or error always renders it a poorer form of evidence. Take for example reporting by the Intergovernmental Panel on Climate Change (IPCC), whose reports regularly include statements about inter alia the observed impacts of climate change, future risks, and the effects of existing climate change policy. Hawley is right that a reductivist need not take these statements as evidence of the beliefs of the IPCC, but they must treat the statements as evidence of some kind, and crucially not as assurance-backed testimony from a trusted scientific organization. Moran’s concern thus arises again: though we think it is sometimes valuable to be able to take an organization at their word, the reductivist cannot accommodate this.

Perhaps the reductivist can insist that IPCC statements are not instances of testimony and should be treated as evidence (of e.g., the impacts of climate change) like any other kind of evidence. But doing so leads them into a dilemma: either they find a way to acknowledge the IPCC’s statements as good evidence, or they deny the epistemic value of those statements. The latter, I think, looks like much too large a bullet to bite. The former looks more promising, but the task of explaining why the IPCC’s statements constitute good evidence looks uncomfortably difficult for the reductivist. If we take this route we do not have recourse to taking the word of the group, so long as we want to reduce testimony to evidence. So instead the reductivist must look for other reasons why these statements can give us reason enough to believe the IPCC.

The reductivist might observe that IPCC reports often include fairly complicated explanations of their findings and the reasons behind their headline conclusions, as well as long lists of citations of evidence that support these findings. These reports do not just ask us to take it on trust. But it is highly unlikely that most readers of these reports will have the time or the expertise to read all of the evidence cited and come to their own conclusions about whether the IPCC’s statements are well-founded. This means that without the option of taking their word for it, many readers of the IPCC reports will not have the capacity to take these reports as reliable evidence.

Hawley’s other reason for rejecting a Moran-style assurance theory for groups is that groups do not freely choose the words of their testimony. But this is false; some groups do choose their words, and more importantly do so in a way that allows us to make a meaningful distinction between information gathered from a group, and things we are told by the group. Earlier I noted that a person can tell us something (e.g., witness testimony), present us with evidence (e.g., CCTV footage), or unintentionally reveal something to us (e.g., through body language). These three categories of learning from another apply also to groups. Say that I am concerned by rumors that a factory in my town is producing a damaging amount of water pollution. To see whether the rumors are credible I might (A) take the company that owns the factory at their word and consult their public statements about their commitments to environmentally-responsible manufacturing. Or I might (B) dig a little deeper into the company’s reports to the relevant environment regulation agency and consult the evidence presented by the company about their pollution levels. Or I might instead (C) test the waste water discharged by the factory for myself, or if I am not able to, organize with other concerned locals to find and fund the expertise to do so.

Each method of learning something about the company’s pollution corresponds to the three ways of learning from another outlined above: C is learning from what the company unintentionally reveals as a by-product of its processes; B is learning from the evidence intentionally presented by company for the scrutiny of others; and A is accepting what the company tells us is the case. And just as learning from the freely chosen words of other people is subject to a particular kind of vulnerability – that the speaker might either intentionally or unintentionally mislead us – so too is A made vulnerable by the fact it depends on words chosen by the company. Moran’s motivation, then, for resisting the reduction of testimony to other forms of evidence can also arise in group cases.

Hawley’s skepticism maintains that there is no evident role for assurance, trustworthiness, or taking responsibility when a group provides testimony. But this argument depends on denying a difference between the statements issued by groups – such as the reports of the IPCC or a company’s public commitments to responsible business practice – and other ways we can source information from groups. In denying this, Hawley neglects the ways in which groups can intentionally mislead and misinform, an epistemic threat that group trustworthiness is fit to defuse.

3. Hawley’s ethical group-trust skepticism

The ethical side to Hawley’s case for reductionism focuses primarily on the reactive attitudes, which have played a significant role in a number of philosophical accounts of the ethical difference between trust and reliance (e.g., Hawley, 2014; Holton, 1994). A reactive-attitude account of the trust/reliance distinction maintains that trust is a morally-loaded subspecies of reliance, distinguished by the fact that when we trust a person they are subject to reactive attitudes toward their response to our trust. Such an account maintains, of course, that a trustee must be a legitimate object of reactive attitudes, but according to Hawley reactive attitudes are not appropriate to groups. She maintains that much of our ostensibly group-directed reactive attitudes (hereafter GDRAs) can be explained away as reactive attitudes directed either toward individual members of groups, or toward social structures that transcend groups (Hawley, 2017, pp. 243–246). Hawley dismisses the viability of ethical trust in groups on these grounds.

What about when we at least appear to blame groups specifically, rather than individual members or society more generally? Some have cited the example of the Deepwater Horizon oil spill in the Gulf of Mexico, where BP’s negligence killed 11 workers and caused devastating environmental damage (French, 2017; Shoemaker, 2019). Widespread anger toward BP in the wake of the disaster seems like a good candidate for a GDRA, because it is anger in response to the blameworthy negligence of an organization. Nonetheless, there are reasons to think that Hawley’s skepticism about GDRAs is warranted even in cases like this. Here I consider two arguments that lend support to Hawley on this matter. This is enough, I suggest, to at least put into question the concept of GDRAs, though (as I will argue shortly) not enough to support her broader group-trust skepticism.

The first argument to support Hawley depends on a claim about reactive attitudes that is made by some opponents of the corporate moral responsibility (CMR) thesis. CMR maintains that corporations like BP can be morally responsible in a way that does not reduce to the moral responsibility of its members (see e.g., Baddorf, 2017; Sepinwall, 2017). According to the relevant CMR opponents, when we hold a reactive attitude we want the object of our attitude to acknowledge the rightness or wrongness of what they have done, and this acknowledgment requires that they empathize with those affected by their actions, feeling the pain of those they have wronged.⁴ This requirement has been posited on two grounds: first, epistemic, insofar as reactive attitudes are only satisfied if their object attains what Baddorf calls “experiential understanding” of the effects of his actions (Baddorf, 2017, p.2781); and second, motivational, insofar as reactive attitudes want their object to be motivated to make amends by empathic suffering (e.g., Shoemaker, 2015, p. 110). These same CMR opponents maintain that because corporations are incapable of phenomenal consciousness they cannot feel anything, which means they cannot be made to feel the pain of their wrongs against others, and therefore cannot be legitimate objects of reactive attitudes (and further that they are incapable of moral responsibility, though this last inference need not concern us here). On this view anger toward BP is not necessarily illegitimate, but it is more like anger toward a malfunctioning machine than indignation toward a person, and it would thereby be a mistake to categorize this anger as a GDRA.

The second argument in favor of Hawley’s GDRA skepticism takes us back to Strawson. Strawson maintained that the reactive attitudes are a way of taking the actions of others personally: “it matters to us whether the actions of other people … reflect attitudes toward us of goodwill, affection, or esteem on the one hand, or contempt, indifference, or malevolence on the other” (Strawson, 1962 [1993], p. 49).⁵ Thus we are offended when another person intentionally harms us, or complimented when another treats us with favor or kindness, and in this respect reactive attitudes register an additional value to interpersonal actions over and above the value of those actions’ consequences for us. More specifically, that additional value lies in what the actions of others say about the way and the degree to which others value me, whether in the form of esteem or a moral respect for personhood. The disanalogy with attitudes toward groups lies in the fact that we are much less often in a position to reasonably think there is anything personal involved in the way that groups, particularly large organizations and corporations, treat us. For instance BP’s negligence might have been caused by collective incompetence, internal dysfunction, or cutting corners in the interest of profit, but none of these explanations for their blameworthy actions involves a personal slight against those who are (legitimately) angry at BP.

More can be said in attempt to rescue GDRAs. Björnsson and Hess (2017) have argued in response to the first argument that corporations are capable of a functional equivalent of guilt that can satisfy a reactive attitude’s intention to produce empathy without the corporation needing phenomenal consciousness. And in response to Hawley, Pouryousefi and Tallant (2022) have cited data on attitudes toward UK banks to argue that anger toward banks, and not bankers, is common, and moreover that this particular anger is a reactive attitude. There is, then, scope for more debate on this issue and I do not propose to conclusively argue against GDRAs. Nor do I wish to deny the legitimacy of the emotional responses that many of us have to corporate misconduct, whether or not we think those responses are accurately categorized as reactive attitudes. I do, however, propose that Hawley’s skepticism on this particular issue has merit, and that there is a burden of proof with those who wish to use the concept of GDRAs, and by extension accounts of group-trust that depend on this concept.

But though I support Hawley’s skepticism about GDRAs, I do not believe this is reason enough to also be skeptical about group trust. This is because there are reasons to think that even if we dismiss GDRAs, there is still room for an ethically significant group trust/reliance distinction. First, we can identify instances of relying on group conduct that are at least prima facie different from trusting groups, and without appealing to reactive attitudes. It is possible, for instance, to rely on groups failing; thus a corrupt politician might rely on the shortcomings of their society’s political system, and perhaps its judicial system, in order to continue to get away with it. But it would be odd to say that such a politician trusts their society’s political and judicial institutions. Moreover, when we talk about the value of trust in these institutions, we do not mean the kind of attitude held by this corrupt politician. It is also possible to rely on institutions to do certain things even if we disapprove of their reasons for doing so. Perhaps I am someone who generally disapproves of pharmaceutical markets populated by firms exclusively motivated by profit, but in the case of the Covid vaccine that profit-motive could be a reliable mechanism for developing a vaccine as quickly as possible. If that is the case, it makes sense to say both that I am happy to rely on pharmaceutical companies for rapid vaccine development and that I do not trust these companies. A distinction seems readily available, without appeal to reactive attitudes, even if we have not yet explained the distinction.

Second, philosophical accounts of trust that make a distinction between trust and reliance are not exhausted by reactive attitude accounts. There are other ways of making this distinction, and reasons to reject a reactive-attitude based group-trust/reliance distinction do not evidently extend to these other theories. Some theorists have maintained a dependency-responsiveness approach, according to which trust is distinguished by its confidence that the person we trust is motivated to act favorably by the fact that I depend on them (e.g., Jones, 2012). Other motivation-based accounts have argued that those we trust are distinguished from those we rely on by being motivated to act favorably by their good will toward the truster (Baier, 1986; Jones, 1996), by their integrity (McLeod, 2002), or by moral obligations (Cogley, 2012). Still others have proposed commitments-based accounts, with Hawley’s own commitments account maintaining that “to trust someone to do something is to believe that she has a commitment to doing it, and to rely upon her to meet that commitment” (Hawley, 2019, p. 9; I will return to commitments accounts in more detail in section 4).

Each of these theories, like reactive attitude theories, take for granted that those we trust are presumed to exercise enough agency to act on the moral psychology that trust ascribes. In this regard it is common among a variety of philosophical accounts of trust to take for granted that trusters presume trustees to be agents in some sense. But it is only reactive attitude accounts that claim further that this agency must be the kind of agency that renders the trustee a legitimate object of reactive attitudes. Reactive attitudes are an indication of morally relevant agency, but we should not let the tail wag the dog on this matter, for an indicator of agency is not a conclusive test of agency. The fact that groups are not legitimate objects of reactive attitudes, if it is indeed a fact, is not sufficient reason to conclude they do not exercise the agency that distinguishes trust from reliance (I will argue that it is indeed the case that groups are capable of a basic form of such agency, without assuming an inflated and controversial metaphysics, in section 5).

4. An interpersonal trust/reliance distinction

I hope to have done enough so far to reclaim conceptual space for a group-trust/reliance distinction. How do we fill that space? In this section I outline an alternative way of making an interpersonal trust/reliance distinction. In the next section I will argue that we can extend this account to group trust.

I will begin my account of an interpersonal trust/reliance distinction by appealing to commitment-based theories of trust, for two reasons. First, Hawley’s own theory of trust is a species of commitment theories (though my commitments account will depart from some of the particulars of Hawley’s), and the closer we come to meeting her skepticism on its own terrain the better chance we have of allaying that skepticism. Second, a commitment account has the advantage of looking elsewhere than reactive attitudes to make a trust/reliance distinction, which means that when we come to extend the commitment account to groups we will avoid presuming too much about the legitimacy of group-directed reactive attitudes (see section 3).

Hawley’s own commitments account maintains that “to trust someone do to something is to believe that she has a commitment to doing it, and to rely upon her to meet that commitment” (Hawley, 2019, p. 9). The appeal to commitments is in part designed to explain cases such as: I rely on, but do not trust, a colleague’s tendency to overestimate how much food they should bring for their lunch, and plan to eat their leftovers rather than bringing my own.⁶ For Hawley, this is not an instance of trust because my colleague has made no commitment to provide for me. Reliance becomes trust if, for instance, my colleague promises to bring my lunch tomorrow.

What makes a person a legitimate object of trust, when they are? On this issue Hawley concurs with reactive attitude accounts, maintaining that unlike at least some forms of relying (on, say, inanimate objects) trustees are treated with a participant stance (Hawley, 2019, p. 14). But for our purposes such a move is best avoided if we want to avoid invoking reactive attitudes. How else might we distinguish trustees from objects of reliance? One option is to maintain that trustees are motivated in ways that objects of reliance are not. Hawley denies that motivations of the trustee are important for trust, on the grounds that popular motivations-based accounts struggle to make sense of what we think of those whom we distrust (Hawley, 2014, pp. 5–6). But even if Hawley is right about distrust, eliminating motivations from our theory of trust leaves us with problematic counterexamples of accidental commitment fulfillment. Say a friend promises to join me and a few others in the pub this weekend. My friend forgets the promise, but she makes the same promise to another friend, who happens to be part of the group joining me in the pub. I can rely on my forgetful friend to be there because she will act as she promised, but she will deliver as promised unintentionally, and I ought not to take this as evidence of her trustworthiness. It is preferable, then, to stipulate that trust expects the trustee to be motivated by the relevant commitment. In light of examples like this, I have argued elsewhere for the following amended commitments-account: “when we trust a person to Φ we are confident that they will be motivated to Φ by a commitment we ascribe to them” (M. Bennett, 2021).

On this view, then, trustees must be agents motivated by a commitment; objects of reliance that are not trustees are not so motivated. But what is a commitment? In this context the term is ambiguous, referring either (or both) to what I call “normative commitment” – that is, a normative demand that I have committed myself to fulfilling – or “psychological commitment” – being driven to fulfil a goal, project, or value to which I am committed (M. Bennett, 2021, p. 528). Commitments accounts sometimes focus exclusively on normative commitments (Hawley’s is an example of this), but some important examples of trust are misrepresented when described in terms of normative demands. Consider: trusting an opponent in a game to be a good sport because of my confidence in their commitment to fair play; trusting a colleague to do a good job, and not just fulfil the minimal requirements of their role; and trust between friends to spend time with and support each other. When we trust in these ways our expectations would be disappointed if the trustee does no more than meet the relevant normative requirements, or if they act as we trust them to out of a sense of duty rather than because they, like us, value those actions.

Expanding the commitments account to include psychological commitments has the virtue of being more comprehensive and capturing cases like these. It also has the virtue of including a category of trust-distinguishing motivations that are several steps removed from reactive attitudes. For if my trust in my colleague is based on their psychological commitment to doing the best job they can, and it turns out that this commitment is not as strong as I hoped, the colleague might let me down but they do not thereby betray me, and feelings of indignation or resentment are inappropriate. In other words, if we take trust to include confidence in others that is not grounded in normative demands, then reactive attitudes need not be part of our account of trust’s distinguishing features.

However, one might worry that by broadening trust in this fashion we have eroded the trust/reliance distinction. Say that I make plans on the basis that I am confident my colleague will do more than minimally required. Above I have referred to this as an instance of trust, but why not call it reliance? The problem arises, I suggest, because thus far the way I have described examples such as the colleague-case underdetermines whether they are examples of trust or reliance. Whether they are cases of trust or reliance depends on further detail about how I take what I know about the psychology of the trustee to give me a reason to be confident about their future conduct.

There are two very different ways I can ground my confidence in that psychology. I might take it as evidence bearing on the probability that in the future the trustee will behave in a certain way. In doing so, the psychology of the colleague is important to me as a factor informing my prediction of what they will do. Perhaps, for instance, I know that my colleague takes pleasure in industry, and yearns for the professional recognition of others. On this basis, I can expect the appropriate incentives to pull my colleague in certain directions that will be favorable to me. Call this a theoretical view of my colleague’s future behavior.

Alternatively, I can have confidence in the combination of the values that my colleague holds and their capacity to act in ways that realize those values. I take my colleague to be someone who thinks our work is important and who sees value in doing that work to the best of our abilities. I am also confident that they have the same understanding as me of what it would mean to do our job well, and that they have the capacity to deliver on this. In short, I have confidence in their practical rationality. Call this a practical view of my colleague’s future behavior.⁷

Reliance that is not specifically trusting-reliance is compatible with the theoretical view because reliance is indiscriminate regarding the motivations of the person on whom we rely. What matters is the likelihood that the relied upon person will act favorably, not why they do so. This is why Kant’s neighbors, so the story goes, could rely on his precise and consistent daily routine to tell the time, without knowing anything about the reasons for his routine behavior. This is also why reliance on Kant could survive learning that his behavior is explained by a pathological compulsion to an orderly schedule. Learning this might lead his neighbors to withdraw reliance for fear of exploiting Kant’s pathology, but this would be because they have ethical concerns about the reliance, and not because they no longer find him a reliable source for telling the time. The latter is threatened only by evidence that undermines our predictions that Kant will continue to act as we need him to. When we attend to such evidence alone, we treat Kant with a theoretical view.

According to the commitments theory I am adopting here, trust is more discriminating, for it grounds its confidence that another will act favorably in the trustee’s practical rationality. Thus if we trust a person because we think they will be motivated by a relevant normative commitment, we are confident that they understand this commitment and have whatever rational capacities are necessary to do what is required of them. And if we trust a person because we think they will be motivated by a relevant psychological commitment, we trust the person to be responsive to reasons generated by their commitment, and to at least try to act in ways that help to realize the values, goals, and projects to which they are committed. Thus on this account when we trust a person we make plans on the supposition that they will exercise their practical rationality to deliver on the commitments, normative or psychological, that support actions favorable to us. In doing so, trust takes the practical view of the trusted, treating them as a practically rational agent, and not as one source of evidence among others about their future conduct.

5. A group trust/reliance distinction

When we rely on someone we are unconcerned with their motivations, because what matters to us is only that they act in a favorable way, not why. By contrast, when we trust a person our confidence that we can rely on them is grounded in their practical rationality, specifically their ability to undertake, recognize, and respond to their commitments and the reasons generated by those commitments. In this section I argue we can distinguish group-trust from group-reliance in an analogous fashion.

Recall that one kind of case that indicates a group trust/reliance distinction is relying on an institution to fail. My example in section 3 was a corrupt politician relying on their dysfunctional political system to let them get away with it. For this politician, their reliance does not depend on any specific understanding of the commitments of the institutions that ought to be preventing political corruption. It could be that the politician has enough power to extort or blackmail their way out of accountability, or that the relevant anti-corruption institutions are themselves run by corrupt bureaucrats, or that these monitoring institutions are simply incompetent. For the corrupt politician, it does not matter; what matters is that they have reason enough to think that they will not be caught, and that they can make plans on this supposition. By contrast, the commitments trust/reliance distinction tells us that if I trust my society’s institutions to prevent such corruption, this trust involves a confidence that the institutions are indeed committed to fulfilling their function, that they have the ability to deliver on this commitment, and that they will respond appropriately to other relevant parties alerting them to the possibility of corruption. Reliance becomes trust when I put my faith in the practical rationality of the anti-corruption institutions (doubts that institutions are capable of exercising practical rationality will be addressed later in this section).

We can draw the distinction in epistemic cases in a similar way. Recall my example in section 2: I wish to learn whether the rumors of a polluting factory in my town are true, and seek information from the company that runs the factory. If I test the factory’s waste water, I rely on evidence unintentionally produced by the company to reach a judgment about the pollution levels. This evidence is subject to a range of epistemic norms, but the company’s trustworthiness is not relevant to the reliability of that evidence. Conversely, if I take the company at its word and form my belief about its pollution on the basis of its public statements, my doing so can be justified if I have good reason to trust the company to tell the truth.

For the commitments account, the latter is distinguished by the fact that my confidence in the company’s public statements can be well grounded in other (also well-grounded) judgments I have made about the company’s commitments to, say, transparency on matters of environmental protection, and their ability to put their commitments into practice effectively. The first means of sourcing information cannot be grounded in this way because the pollutant levels in the waste water are unintentional, and do not reflect anything about the commitments of the company or its ability to live up to those commitments. By contrast, if I test the rumors about pollution by consulting the statements issued by the company, I must be relevantly confident in the agency of the company because they have control over the words in the statement.

One might object that the distinction between trust and reliance in these cases can be understood without a concept of group trust. It could be that when I take a company at its word, or when I am confident in the favorable conduct of an institution because of its commitments, what really grounds my confidence are the commitments of individual group-members and not anything about the group per se. The relevant distinction, then, would not be between group-reliance and group-trust, but rather between group-reliance and individual-member-trust.

The problem with this objection is that individual-member commitments will not always explain trust in groups. I can, for instance, be confident in an institution’s commitments despite a change of personnel, or despite not knowing anything about the individuals running the institution. Less happily, I can also judge that an institution is unlikely to deliver on its commitments despite having confidence in its individual members. Large organizations with complex bureaucracy might be staffed and run with individuals who I think have all the right qualities to be relevantly trustworthy, but I might nonetheless think these individuals will be inevitably thwarted by the internal dysfunction of the organization they work in. Thus my judgments about a group’s capacity to undertake, recognize, and respond to relevant commitments can diverge from analogous judgments about the group’s members, and the former do not reduce to the latter in all cases.

Alternatively one might object that this account of the group trust/reliance distinction depends on the contested assumption that groups are capable of being agents. This objection can be rebutted, I believe, but the notion of group agency that my account depends on certainly needs to be unpacked and defended. That agency must be minimal if we are to avoid some of the more controversial topics in discussions of group agency.⁸ Luckily, we need not set the agency-requirements for groups especially high in order to apply the trust/reliance distinction.

The commitments account I have adopted requires that an entity possesses three agential capacities if it is to be considered a legitimate object of trust. First (A) a trustee must be able to make commitments, for if it cannot then it cannot be motivated by commitments, and the distinctive motivational ascription of trust (according to the commitments theory) would be unavailable. But undertaking commitments cannot be enough to ground the confidence of trust, for we might think that a potential trustee can undertake a commitment but cannot recognize and act on it. Hence an object of trust must also (B) be capable of delivering on its commitments. The capacity to deliver on psychological commitments requires inter alia reflexive awareness (“what do I value?”) and the practical reasoning to combine this self-awareness with information about my environment in order to realize my commitments. The capacity to deliver on normative commitments requires inter alia the ability to understand the relevant norms, how they apply to me, and what I ought to do to meet those norms given the commitments I have undertaken.

The rational capacities of trustees must also include (C) a capacity to respond to reasons provided by others. This is because, while the purest form of trust will completely defer to the trustee and leave it up to them to figure out for themselves how best to deliver on the relevant commitment, most of the time trusting relationships are less than wholly pure and still involve a truster reasoning with the trustee. Thus when I trust in a colleague to do a good job I am likely to sometimes leave it wholly up to them to take care of relevant matters, but might also work with them, share information and know-how, or discuss the best approach to our work collaboratively, on the understanding that my colleague does indeed want to do a good job and will incorporate our conversations into their reasoning from commitment to action. Trust that does not defer and micro-manages is no trust at all, but trust need not in all cases be a complete outsourcing of reasoning to the trustee.

These basic practical rational capacities do not usually feature explicitly in accounts of trust because they can largely be taken for granted when we are discussing trusting other people. But what about groups? Regarding A, groups can and often do undertake both normative and psychological commitments. Recall section 2’s factory, rumored to be polluting surrounding bodies of water. Such factories are governed by laws and regulations that determine the appropriate level of water pollutants they may produce. Accordingly, companies that own and run such factories undertake normative commitments to abide by those regulations simply by choosing to operate the factory and expel waste water. More generally, very many groups among organizations, public institutions, and businesses will at least sometimes be parties to contracts. Wherever this happens, the group undertakes a normative commitment.

The term “psychological commitment” does not seem as promising for an uncontroversial account of group agential capacities. This term might, for instance, be taken to presume mental states that cannot be taken for granted when discussing groups. But we should not be misled by terminology. Psychological commitments differ from normative commitments insofar as they depend on the trustee being motivated by the relevant value, project, goal etc. This is what we mean when we say that someone is committed to, say, a political cause. An analogous kind of commitment can be meaningfully ascribed to groups without invoking group mental states. Many of us might be most familiar with only bad examples of such commitments among groups, as when large corporations claim to uphold certain socially responsible values in order to launder their public image. But that such groups can intelligibly claim such values at all indicates something more general: that businesses, organizations, and public institutions can be committed to upholding certain values.

Moreover if we accept that groups can undertake commitments, then there seems to be no reason to deny that groups sometimes can (B) understand these commitments, combine them with information about their environment, and act in an effort to meet or realize their commitments. Thus factory-owning companies are generally taken to be capable of understanding their legal obligations. And where, say, an employer claims to be committed to equitable wages, we would usually expect the employer to have some understanding of what this entails and to take measures to deliver on this commitment.

What about (C), the capacity to respond to reasons? It might be that opportunities to reason with groups are less common than opportunities to reason with people we trust, but they exist. Being a trade union member, for instance, can sometimes involve a somewhat messy mix of, on the one hand, trusting the union to collectively reason its way to workable strategies, and on the other hand actively engaging in collective discussions, making the case for a position that not all members agree with, or drawing attention to concerns aligned with the commitments of the union that would otherwise be overlooked. Complete deferral is not always an attractive position for a union member, and it would be too demanding to insist that trusting the union rules out reasoning with other members or attempting to influence the union in order to better achieve its goals. Trust in groups, in situations like this, also involves not just taking for granted the group’s capacity to make and act on commitments, but also the group’s responsiveness to reasoning from others in order to guide the group to better act on their commitments. And as membership of unions sometimes illustrates, we do indeed sometimes relate to groups on the understanding that they possess this responsiveness.

6. The value of a group trust/reliance distinction

I submit, then, that the commitments-based distinction between trust and reliance can be made with regard to groups. But we might still wonder whether this distinction, available in theory, helps us understand anything about our practice of trusting groups. To borrow a phrase from Karen Jones, we might ask how much “conceptual work” is done by the distinction I have argued for (Jones, 2017, p. 99).

By my count, there are two benefits to the distinction I have argued for. First, it can help us further understand why trust, when well-placed, can be more valuable than confidence secured through monitoring and accountability mechanisms. As both Onora O’Neill and C. Thi Nguyen have argued, when we use transparency mechanisms to ensure an institution does its job to our satisfaction, we can damage the conduct of that institution, through incentivizing dishonesty (O’Neill, 2002) or disincentivising valuable activity that does not lend itself well to public justification (Nguyen, 2021). But there is additional value to group-trust even when the activity of an institution would be the same whether achieved through monitoring or through trusting. That is, it is better that the same conduct is delivered on trust than through force of accountability. This additional value is evident, for instance, in the expressive function of declaring trust in a political party; we can express our support of a party by declaring our trust in it, but we cannot express this support as effectively by declaring reliance. Even stating that a party is reliable does not show support as directly as stating that we trust them.

The commitments account gives us a way of parsing this additional value of group-trust. There is value in knowing that the institution, in addition to acting favorably, has favorable motivations, commitments that explain its activity and that I endorse. I prefer, for instance, that a political party that campaigns on a manifesto I endorse enacts that policy in government because it is committed to it – it reflects what the party in fact values and stands for – rather than because the party fears public backlash. Electoral accountability might ensure the party keeps its promise, but if this is what grounds my confidence in the party then I am without the additional value that trust brings because I am not also confident in the party’s commitments. That confidence in their commitments can also secure an additional expressive function for declaring trust, for when I declare my trust in the party I signal that their motivating values align with my own. Through declaring trust, I can imply that the party is, as it were, on my side, because trusting entails a favorable judgment of the party’s principles.

The second benefit to the trust/reliance distinction I have argued for is that it vindicates some instances of social scientists using the terms “trust” and “trustworthiness” when studying public confidence in institutions. Consider for example the decades long Trust in Government index from the American National Election Study (ANES), which shows a long-term decline in American trust in government (American National Election Study, 2020). A skeptic about group trust will suspect that what the study is really measuring is neither trust nor judgments of government trustworthiness, but judgments of reliability. But the commitments trust/reliance distinction rescues the Trust in Government index from this skepticism.

The ANES trust score is based on a small number of questions, one of which is a question about whether the government is wasteful with tax revenue. Such a question solicits opinions about the efficiency and competence of government, with no distinctively trust-relevant judgments involved. Were this the only basis of the score, the commitments trust/reliance distinction would have to concede that the ANES trust index is really a reliability index. But the ANES trust score is also based on questions that go beyond judgments of government competence, including the crucial question: “Would you say the government is pretty much run by a few big interests looking out for themselves or that it is run for the benefit of all the people?” The ANES trust score is thus partly based on respondent judgments not just about the functioning of government, nor even just about government acting favorably, but about whose interests motivate the actions of the government, whose benefit the government works for, or, in the terms of the account I have argued for, what the government is committed to. In this respect, the commitments trust/reliance distinction helps us see why the ANES measure merits the term “trust”.⁹

7. Conclusion

Hawley’s skepticism about group trust raises important questions about whether the trust/reliance distinction, so significant for philosophies of interpersonal trust, can extend to our attitudes toward groups. Pace Hawley, I have argued that we can make good on an intuition pumped by some cases of group trust that there is indeed a difference between trusting and relying on groups. We can rely on groups to fail, and we can rely on groups to do the right things for the wrong reasons, but trust is a concept that seems ill suited to these cases. We can also rely on information sourced from groups, but we reserve the concept of trusting the testimony of a group for when we take a group’s word for it.

The group trust/reliance difference can be parsed through appeal to a commitments account of trust, according to which trust is a distinctive kind of reliance that requires we take a practical stance toward the trustee and ground our confidence in their conduct in a positive appraisal of their practical rationality, foremost the kinds of commitments we ascribe to them and their ability to act on those commitments accordingly. This is one way we can make sense of the difference between trusting and relying on individual people. I have argued it is also a way we can make sense of the difference between trusting and relying on groups.

Disclosure statement

No potential conflict of interest was reported by the author.

Additional information

Funding

The work was supported by the The Leverhulme Trust [Research Leadership Award, University of Essex].

Notes

1. One recent exception is Pouryousefi and Tallant (2022), who argue that a range of empirical data from trust research puts significant pressure on Hawley’s group-trust skepticism. In this paper I reach the same conclusion via a different route (conceptual argumentation) and attempt not just to show that group trust is a distinctive phenomenon, but to also give my own theoretical account of a group trust/reliance distinction.

2. See e.g., American National Election Study (2020); R. Bennett and Kottasz (2012); and O’Neill (2002). Norris (2022) also maintains that we suffer from crises of excessive public trust (credulity) in some contexts.

3. Dormandy (2019, p. 14) helpfully situates trust-based views in the broader landscape of testimony epistemology.

4. And the happiness of those they have helped, though the reactive attitudes literature typically focuses exclusively on wrongdoers.

5. This feature of Strawson’s account is reflected in more recent reactive attitude theories that continue to characterize these attitudes as responses to a “slight”. See e.g., Shoemaker (2015), p.110.

6. Following Holton (1994), I take relying to be planning on the supposition that something will happen.

7. A similar distinction between theoretical and practical views, and the relevance of this distinction to trust, has been defended at greater length by Marušić (2015, especially chapter 7).

8. For example: must a group agent have a group mind (Silver, 2022)? Do group agents have free will (Hess, 2014)? Are groups capable of moral responsibility (Gilbert, 2006)?

9. Trust measures that include questions soliciting beliefs about an institution’s motivating values, interests, or commitments are also used in studies of trust in banks (e.g., R. Bennett & Kottasz, 2012) and trust in politics (e.g., Valgarðsson et al., 2021).