Towards a Non-Reliance Commitment Account of Trust

1 Introduction

Trust is commonly taken by philosophers to be a metaphysically-hybrid notion involving an attitude and an action (Faulkner 2015; Hawley 2012). The action component of trust is typically defined as a special form of reliance in which the trustor has: (1) heightened expectations of their trustee; and (2) a disposition to justifiably feel betrayed if their trust is broken by their trustee (Baier 1986; Jones 2004; Hawley 2012; McLeod 2015). The first aim of this paper is to reject this trust-as-reliance intuition.

The second aim of this paper is to develop a non-reliance account of trust. Following from my previous argument that a commitment account of trust is the best candidate for an adequate theory because it captures more cases of trust (Kelsall 2022), I develop a commitment account in this paper.¹ According to Hawley’s commitment account, X trusts Y only if X believes Y to have a commitment to ϕ and relies upon Y to keep that commitment to ϕ (Hawley, 2019: 9). Because Hawley’s commitment account is reliance-based, I propose a non-reliance-based commitment account. Roughly, on my account, if X trusts Y, then X has a belief with the following counterfactual as its content: ‘if I, X were to rely on Y to keep a commitment to ϕ, then I believe that Y would keep that commitment to ϕ.’

In Sect. 2, I explain and motivate the trust-as-reliance intuition. In Sect. 3, I present four counterexamples to the reliance-based trust accounts. I use one of these counterexamples to also argue that if trust is reliance, it cannot explain the process of reconciliation by which we come to trust trustees again after betrayal. In Sect. 4, I introduce my non-reliance-based commitment account. I then defend it against the objections that it cannot make sense of the heightened expectations and connections to betrayal that are commonly accepted as features of trust.

2 The Trust-as-Reliance Intuition

In Sect. 2.1, I explain the intuition that trust is a special form of reliance that is distinct from mere reliance. In Sect. 2.2, I show how the trust-as-reliance intuition arises in two general accounts of trust: motive-based and non-motive-based accounts. Then, in Sect. 2.3, I present the strongest arguments for the trust-as-reliance intuition which are: (1) that it is required to explain the necessary connection between trust and betrayal; and (2) that it is required to explain the existence of so-called therapeutic trust.

2.1 Distinguishing Trust from Mere Reliance

Trust and reliance are used synonymously in ordinary parlance, so any attempt to disambiguate them seems wrong from the start. Nevertheless, even those who believe that trust is reliance distinguish it as a special form of reliance. They argue that trust involves two things that are absent in other forms of reliance: (1) the trustor’s heightened expectations of the one trusted; and (2) a propensity to feel betrayal if said expectations are not met (Baier 1986; Jones 2004; Hawley 2012; McLeod 2015).

Defining the specific content of the heightened expectation of trustors varies between different trust accounts, but at the highest level of generality it is the normative expectation that trustees will make good on trust at least in part because they are being relied upon to do so (McGeer & Pettit 2017). This is different from the predictive expectation one has when relying on objects to fulfil their purpose, such as predicting that the car will get us to work. Of course, one can have predictive expectations of a trustee as well; it would be odd to trust someone to watch your bag if you didn’t also think that they would watch it. However, trust involves these heightened normative expectations alongside the predictive expectations that occur in cases of general reliance (Margalit 2017: 142).

It is commonly accepted that when trust fails, the failure elicits justified feelings of resentment and betrayal from the trustor towards the trustee, whereas mere reliance elicits only disappointment (Baier 1986; Hawley 2012; McLeod 2015). This is correct. Imagine seeing someone in front of a broken-down car throwing his hands up in the air and declaring, ‘you have betrayed me!’ Even in the absence of this dramatic display, the idea that a car can betray you by breaking down is absurd. One can be reasonably frustrated at the situation, that is, disappointed, but betrayal is a stretch too far. It’s not like an inanimate object can wrong you in any moral sense, as much as it might frustrate you by breaking down. On the other hand, one could justifiably feel betrayed by the dodgy salesman who deceived you into buying the car. This is because reliance on the salesman involves the justified normative expectation that he will be honest and sincere when telling you about the car, an expectation he breaks by deceiving you.²

That betrayal is linked to trust and not mere reliance makes sense when we consider it in conjunction with trustors’ heightened expectations of trustees. If, in trusting, we expect trustees to be motivated to make good on our trust at least partly in virtue of our dependence on them, then the breaking of trust reveals to us that the trustee does not value or care about our dependence on them for its own sake. Thus, Margalit says of betrayal:

‘The issue in betrayal is not the intention to harm but the indifference to the victim, the lack of concern. The true insult in betrayal is the discovery by the betrayed that he wasn’t at all on the mind of the betrayer. Indeed, what was done was not directed against him but instead was done with utter disregard for him.’ (Margalit 2017: 112).

The harm of betrayal that Margalit describes is distinct from the trustor’s losing out on the object of trust. Suppose Sally trusts Joanne with a dark secret. Joanne tells Arabella the secret, but on her way to spread it about town, Arabella is killed in a car accident. In this case, Sally doesn’t suffer any consequences with respect to the object of trust as the secret doesn’t get out. Nevertheless, Joanne has betrayed her because she still disregarded Sally’s dependence on her by revealing the secret.

2.2 Theories of Trust

So far, we’ve defined trust as a special kind of reliance involving heightened normative expectations and a propensity to feel justifiably betrayed when let down. Before arguing against the trust-as-reliance intuition, it will be useful to see how the intuition manifests across two kinds of trust accounts: motive-based accounts and non-motive-based accounts.

On motives-based accounts, the heightened expectations of trustors are not merely normative expectations that the trustor will do as trusted, but that they will do as trusted in virtue of specific motivations. The most prominent account of this kind is Baier’s goodwill account. Roughly, Baier holds that trust is a special form of reliance in which the trustor relies upon the goodwill of the trustee (Baier 1986: 234-5; 1992).

However, having a motives-based account of trust doesn’t commit one to the trust-as-reliance intuition. On Hardin’s encapsulated interest theory of trust, to trust is not to rely, but to judge that in some instance of reliance, the trustee will encapsulate your interests among their own when determining how to act, such that they will satisfy your reliance (Hardin 2006: 19). So, for Baier, to trust is to rely on a person to act with a specific motivation (goodwill), while for Hardin, to trust is to judge that if one were to rely, then the trustee would be motivated to encapsulate your interests among their own enough to do as trusted.

For supporters of motives-based accounts, specific motivations are necessary to make sense of the betrayal associated with failed trust. The thought goes that if X relies on Y to tell the truth via a coercive method, then X isn’t really trusting Y, but rather, depending on Y’s susceptibility to coercion rather than trustworthiness (Baier 1992; Jones 1996). In such a case, if Y fails to tell the truth, there can only be disappointment, not betrayal, because trust requires reliance on the trustee’s willingness to voluntarily choose to do as trusted, (through good-will or encapsulated interest, or some other motivation).³ However, recent work by Hawley (2019) and myself (Kelsall 2022) has called into question whether reliance specific motivations is necessary for trust. I discuss this further in Sect. 4; for now, I present Hawley’s commitment account as an example of a non-motives-based, reliance account of trust.

On Hawley’s commitment account, the heightened expectation of the trustor is the normative expectation that the trustee keeps their commitment; whatever motivation a trustee may have for making the commitment is irrelevant. A person takes up a commitment when they either tacitly or implicitly invite others to depend upon them to some end. An example of an explicit commitment would be promise making (Hawley 2019: chp.2). An example of implicit commitment making would be a commitment to honesty if one answers a question, such as ‘are there any beers in the fridge?’ The thought here is that in answering the question, the norms of discourse imply that you are responding truthfully to the question, even if you don’t explicitly state the intention to speak truthfully (Hawley 2019: chp.3).

Here are Hawley’s definitions of trust and distrust.

‘Trust – 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.

Distrust – To distrust someone to do something is to believe that she has a commitment to doing it, and yet not rely upon her to meet that commitment’ (Hawley 2019: 9).

Hawley’s account is reliance-based. The trustor must rely on the trustee to keep a commitment, as well as believe that they have that commitment. On the alternative account I propose in Sect. 4, to trust will be to believe that if one were to rely on a trustee in a special kind of way (I call this trusting reliance) then one trusts that trustee. So, the account I propose is an example of a non-reliance, non-motivation, commitment trust account.⁴

2.3 Why Accept the Trust-as-Reliance Intuition?

Now we know how the trust-as-reliance intuition works in different theories of trust, we should consider the strongest arguments for it. I do so by explaining two objections to non-reliance accounts of trust, such as Hardin’s and Hieronymi’s, for whom trust is a matter of belief alone.

For Hieronymi and Hardin, trust requires the trustor to believe that their trustees are trustworthy (Hardin 2006: 17; Hieronymi 2008: 214).⁵ Hieronymi even calls belief accounts of trust ‘purists’ notions of trust (ibid). These accounts are non-reliance accounts because all they require is that trustors take a certain attitude towards the trustee, such as a belief in good intentions and competence to do as trusted, to count as trusting. Thus, we may also call them attitude-based instead of non-reliance-based. My own theory is attitude-based, since trust also amounts to the trustor having a belief about the trustworthiness of a trustee, although trustworthiness on my account is cast in terms of commitment.

Critics of attitude-based accounts argue that if trust is a matter of belief alone, then we cannot account for a special form of trust, namely therapeutic trust (McGeer 2008). Therapeutic trust was first discussed by Horsburgh (1960), who describes it as occurring when a trustor appears to trust a trustee despite not believing that the trustee is trustworthy. Here, trust is conceived of as reliance in the face of doubts, typically in the hope that the trustee will prove trustworthy.⁶ In placing therapeutic trust, you signal to the trustee that they are worthy of trust, and this gives them extra reason to live up to that trust. Since there is no belief that the trustee is trustworthy yet, only reliance, it seems both that trust isn’t a belief in trustworthiness and that it is a form of reliance.

For McGeer, therapeutic trust is really deserving of the purist title that Hieronymi gives to belief trust. In therapeutic trust, the fact that the trustor relies on the trustee despite the lack of belief that they are trustworthy, but merely hoping that they will be, appears to be more trusting than the person who confidently trusts someone who they strongly believe to be trustworthy. McGeer argues that therapeutic trust is ‘substantial trust’ (McGeer 2008: 237). For her, it is ultimately what we are willing to do, namely, to rely even in the fact of extreme doubt, that defines trust, rather than what we might believe about our trustees’ trustworthiness (2008: 421).

If we think therapeutic trust is trust, this is a problem for attitude accounts such as Hardin’s, Hieronymi’s, and my commitment account. Therapeutic trust is a counterexample to these accounts, since it is a case where the trustor does not believe that the trustee is trustworthy yet trusts them anyway.

If we want a satisfactory non-reliance trust account where trust is an attitude, then we must address McGeer’s objection from therapeutic trust. I deal with McGeer’s therapeutic trust objection in Sect. 3. The problems don’t end here, however. I must ensure my account explains the intuitions that trust involves heightened expectations and specify its connection to betrayal. The latter is especially important, since another objection against attitude accounts is that they cannot explain the link between trust and betrayal because one cannot be betrayed if one only has a belief but does not rely on anyone. I turn to this in Sect. 4.

3 Rejecting the Reliance Intuition

The aim of this section is to reject the reliance intuition and to respond to McGeer’s therapeutic trust objection. To do this, in Sect. 3.1, I distinguish trust, which I believe is attitude-based, from trusting reliance, which is the kind of reliance that supporters of reliance accounts (wrongly), describe as trust. Then, in Sect.3.2, I present three counterexamples to the trust-as reliance intuition. Finally, in Sect. 3.3, I present what I call the reconciliation argument, which shows how reliance-based accounts struggle to explain how trust is regained after it has been betrayed. With this argument, I respond to McGeer’s view that attitude accounts cannot accommodate therapeutic trust.

3.1 Trust and Trusting Reliance

My counterexamples and reconciliation argument rest on a distinction between trust and what I call trusting reliance. According to this distinction, to trust is to believe that a trustee is fit for trusting reliance with respect to some object or objects of trust. For example, on a goodwill account, to trust would be to believe that a trustee will be motivated by goodwill to make good on reliance in some situation. Trusting reliance is the special form of reliance that conforms to the trusting attitude. Therefore, to trustingly rely on a goodwill account is rely on a trustee’s goodwill.

One might think this distinction is merely splitting hairs, but the arguments in the following sub-sections show that the distinction is significant. Moreover, it accounts for what Domenicucci and Holton (2017) have described as the logic of trust. They note that while it is true that we colloquially use trust and reliance synonymously, there is a distinctive use of the verb ‘trust’ in which trust is used to justify reliance, in a way that would be nonsensical if applied to the verb ‘reliance’ (Domenicucci and Holton 2017: 152). Consider the following two sentences:

• (1) Ok, I trust you. You can look after Tibby the cat.

• (2) Ok, I rely on you. You can look after Tibby the cat.

If someone uttered (2), we might think they were not a native speaker of English; however, sentences like (1) are acceptable and, Domenicucci and Holton argue, demonstrative of the idea that the question of trust can be separated from and comes prior to the question of reliance. They claim that (1) works in a radically different way from (2). Our talk of trust is used to explain or justify our reliance, which is looking after Tibby. It’s as though the trustor said: “I trust you, therefore you can look after Tibby” (Domenicucci and Holton 2017: 152). Putting these thoughts in my terminology, we form the attitude of trust which then justifies our trusting reliance in trustees.

The distinction also accounts for situations in which we may choose to place trusting reliance in a person, despite lacking the prior belief in their trustworthiness, and situations in which we don’t place trusting reliance in a person, despite our trusting them. The examples in the following subsections will show how this is possible.

3.2 Three Counterexamples to Reliance-based Trust

In this section, I present three counterexamples to the claim that trust requires reliance. Here is the first counterexample to reliance-based trust.

Cat sitter – Sarah needs a cat sitter while she is out of town. Her friends Tom and Anna are both cat lovers, and Sarah believes them to be equally competent and willing to do the job. However, she asks Tom rather than Anna, because he lives on her street whereas Anna lives in the next town.

On a reliance-based account, we are committed to the claim that Sarah trusts Tom but not Anna, since she relies on the former but not the latter. Despite this, Sarah has the same attitude regarding Anna and Tom with respect to their trustworthiness; the only reason she relies on Tom instead of Anna is convenience. It seems wrong to say that Sarah doesn’t trust Anna when: (1) she believes her to be equally trustworthy as Tom; (2) would rely on her if she lived closer; and (3) her reason for not relying on Anna has nothing to do with Anna’s character, but the external factor of convenience. The fact that Sarah is willing to rely on Anna if Tom were not available or if she lived closer, is a reason to think that she trusts Anna just as much as Tom. Moreover, Sarah could provide the following justification to Anna if Sarah complained that she wasn’t trusted: ‘it’s not that I don’t trust you Anna, I do, it’s just more convenient for me to ask Tom since he’s closer.’ What matters for trust is not whether Sarah relies on Tom or Anna, but on what she believes about how her friends would act if she were to rely on them.

Explaining cat sitter with the distinction between trust and trusting reliance, Sarah trusts both her friends, but only trustingly relies on Tom. The reason why Anna might complain of not being trusted is because she takes the lack of trusting reliance to indicate a lack of trust, a worry which Sarah can allay by pointing to the fact that she would trustingly rely on Anna if she were in Tom’s place. Anna’s worry that Sarah doesn’t trust her is understandable. Since to trust is to believe that someone is apt for trusting reliance, the fact that one isn’t trustingly relied upon in what seems to be an appropriate situation where one should be indicates a lack of trust. While it is reasonable for Anna to think this, all Sarah needs to do is point out that she would trustingly rely on her if it were more convenient to do so, and this indicates that she trusts Anna, even if she doesn’t trustingly rely on her in this case.

One might object that I am confusing two senses of trust in my analysis of cat sitter, namely three-place and two-place trust.⁷ Three-place trust is the kind of trust we’ve discussed throughout this paper: ‘X trusts Y to Φ’. Two-place trust occurs when a trustor simply trusts a trustee as in ‘X trusts Y’, but with respect to no special object of trust Φ. Two-place trust can be understood in two ways. The first way is that two-place trust is simply a shorthand way of describing multiple three-place trust relations (Hardin 2002; Hawley 2014: 16). Here, to say that X trusts Y is just to say that there are many objects of trust with which X would trust Y. The second way to understand two-place trust is as something that is not reducible to an object of trust at all. Here, we would say that X is just trusting of Y (Faulkner 2015; Carter & Simon 2020: 5). Although hard to fathom what it could mean to simply trust someone simpliciter with respect to no specific objects of trust, we can think of cases of close relationships, such as familial relationships, friendship, and love, in which trust in each other seems wholly general and not specifiable in three-place terms.

However, I do not think there is a confusion between these two-forms of trust going on here. We can make sense of cat sitter with the distinction between trust and trusting reliance. To see this, consider a modified version of cat sitter.

Enemy vs Friend Cat sitter – Sarah needs a cat sitter while she is out of town. Sarah is best friends with Tom and mortal enemies with Anna. However, Sarah knows that Anna loves cats so much that she would make an equally trustworthy cat sitter as Tom, treating her cat no differently than Tom would, and maintaining a temporary truce between herself and Sarah in the process.

In this case, Sarah doesn’t have two-place trust in Anna in either of the senses described above. She doesn’t have two-place trust in the reductive sense, because the only time she would trust Sarah is as a cat sitter, which is just one instance of three-place trust. On the second interpretation, the fact that they are mortal enemies means that they do not have the kind of generalised trust we might find in close interpersonal relationships. Accepting two-place trust, we ought to say that Sarah has two-place distrust in Anna but does trust her in the three-place sense as a cat sitter.

In both cat sitter and enemy cat sitter, the fact that Sarah believes that were she to trustingly rely on Anna as a cat sitter, Anna would make good on that reliance, shows that Sarah trusts Anna as a cat sitter. She doesn’t trustingly rely on Anna for the external factor of convenience, but she does trust her. Let’s look at the third  counterexample, which further demonstrates how external factors can prevent our placing trusting reliance in trustees despite trusting them.

Regulation: Organisation A is responsible for regulating organisations B, C, and D.⁸ Whatever regulatory approaches A takes towards B, C, and D, it must use the same regulatory processes for each. While B and C are trustworthy organisations in need of little regulation, D is untrustworthy, and requires closer scrutiny. As such, A imposes tight regulations on all three organisations, as consistent with its obligation to equal treatment.

On a reliance account of trust, A trusts none of the organisations since coercive regulation takes the place of the special kind of reliance that defines trust. In my view, although A doesn’t trust D, the fact that A believes B and C to be trustworthy suggests that A trusts B and C. A can explain to B and C that they are trusted but because of the rules about equal regulation and D’s untrustworthiness, A is forced to regulate all three organisations. However, since A believes that B and C would make good on trusting reliance if the regulations were removed, A trusts B and C, even if it can’t trustingly rely on them. Again, external factors prevent an agent from acting out their trust in trusting reliance. A’s regulative practices have nothing to do with B or C, and everything to do with D, and the rules of regulation. Therefore, A trusts B and C, it distrusts D, and it trustingly relies on none of them.

Pettit (1995) argues for the impossibility of trust in cases like regulation, as does O’Neill in her criticism of audit culture (2002). The thought is that because trust is a form of reliance where trustors accept vulnerability to betrayal, attempts to regulate or coerce trustees into doing as trusted are incompatible with it. If we accept the distinction between trust and trusting reliance, however, regulation is only incompatible with trusting reliance, not trust. Of course, as with cat sitter, it can be difficult to convince someone that you trust them where you don’t trustingly rely, but this can be avoided by explaining that you believe that, were it possible or appropriate to trustingly rely on them, then you would do so because you believe they would make good on it. In cat sitter, Sarah can do this to Anna by saying that she would trustingly rely if it were more convenient; in regulation, A can do it by saying that it would trustingly rely on B and C if not for D’s untrustworthiness and the regulatory policy. So long as the trustor believes that the trustee would satisfy trusting reliance, they trust.

3.3 The Reconciliation Argument

In this section, I explain the reconciliation argument against the reliance intuition. It is the argument that if trust is defined as a special form of reliance, we cannot make sense of the process of reconciliation after a betrayal. The argument also demonstrates that therapeutic trust is not a counterexample to non-reliance accounts, because therapeutic trust is not really trust, but trusting reliance.

Consider the following example of an attempt of reconciliation after betrayal:

Cheating: Sam’s partner Susan has cheated on him. However, both decide that they want their relationship to continue. Sam has significant doubts at first whether Susan will live up to her promises; nevertheless, he refrains from checking up on her, recognising that to do so would not allow their relationship to get back to how it was.

Cheating is an example of therapeutic trust. Sam doubts his wife’s trustworthiness, but he relies on her in the hope she will prove trustworthy. As we saw in Sect. 2, McGeer argues that therapeutic trust is not only trust, but substantial trust. The fact that Sam continues to rely on Susan despite his doubts proves that he trusts her. However, in my view it is wrong to say that Sam trusts Susan while he harbours these significant doubts. I would describe cheating as a scenario in which Sam trustingly relies on Susan but doesn’t trust her because he doesn’t believe she is trustworthy. Nevertheless, in placing his trusting reliance in her he hopes she will prove trustworthy and that if she does, he will come to trust her again.

I think my explanation better describes the process of reconciliation towards trust again after a betrayal than McGeer’s. In the wake of her betrayal, Sam distrusts Susan. Susan expresses remorse to Sam. This convinces Sam to give his wife a chance, and so he trustingly relies on her. Yet, in his heart, he does not trust Susan yet because he constantly doubts her fidelity. Perhaps he trustingly relies because he hopes she will be trustworthy, hopes to save his marriage, or is afraid of loneliness. But suppose that with time, Susan makes good on his trusting reliance and, slowly, Sam takes a more positive judgement. Once he reaches the point where he believes that his wife is and will continue to satisfy his trusting reliance, he trusts her again. Until that point, we would say that Sam acts as if he trusts Susan (by trustingly relying on her) to give her the opportunity to prove herself trustworthy and thus regain his trust. McGeer’s notion places the cart before the horse, since Sam trusts again the moment he places his trusting reliance, despite his significant doubts. In fact, Sam trusts the most at the very beginning when he has the most doubt on McGeer’s account, because she describes this situation as one of substantial trust. This seems implausible with how trust is typically seen to be weakest in the immediate aftermath of a betrayal and is something that is slowly rebuilt over time as doubts decrease.

My interpretation of cheating so far has rested on a priori argument and at least partially on intuition. However, my interpretation can be supported by considering psychological studies on the process of reconciliation as articulated and felt by victims of infidelity. First, victims often claim that their reasons for choosing to initially recommit to a relationship are not trust in their partner, but other factors, such as the desire to maintain a family (especially where there are children), spiritual commitments, or a hope that in the fullness of time, they will be able to forgive their partner again (Mitchell et al. 2022; Fife, Weeks & Stelberg-Filbet 2013). In such cases, then, even if partners trustingly rely on their partners by not checking up on them or monitoring them invasively, they still do not trust their partners, since their trusting reliance has nothing or little to do with beliefs about their partner’s trustworthiness, but rather, external desires such as maintaining a family, assets, spirituality, or a hope that things might improve in the future. It seems that the best explanation here is not that the victim trusts, but that they trustingly rely for external reasons such as protecting their family, maintaining spiritual integrity, or perhaps because they also hope to trust again in the future.

Second, victims of infidelity also describe a suspicion about their partners which manifest in extreme urges to monitor them, for example, by having access to their phones, passwords, or geographical location. This happens most strongly in the immediate aftermath of the cheating but even long after the cheating occurred (Mitchell et al. 2022: 454). This need is used as a metric for victims to explain how much they trust their partners; where it is so strong that partners cannot resist trying to actualise it, their trust is weakest, but even when it manifests only in doubts and suspicion, periodically, years later, victims still describe this as showing that their trust in their partners is no longer as deep as it was (Walravens & Rober 2022: 7). If the presence of doubts about a partner’s fidelity are used as a metric for whether one trusts one’s partner, then this gives further support to the claim that trust is about belief in trustworthiness, not reliance. It also supports the claim that the process of reconciliation is one that begins with trusting reliance and not trust, with trust only developing later once the betrayer demonstrates their trustworthiness by satisfying their partner’s trusting reliance, and where this results in the betrayed doubting their partner less and less.

My interpretation of cheating also supports Hieronymi’s claim that betrayers can sometimes legitimately complain to the wronged party that they aren’t fully trusted in cases where the wronged party trustingly relies on but lacks the belief in trustworthiness (2008: 230). Imagine that in cheating Susan discovers how Sam feels about her; she may feel hurt that despite her efforts to prove trustworthy, he still does not trust her in his heart, even if he keeps that distrust to himself.

In a recent paper, Carter argues against Hieronymi by suggesting that we could equally conceive of the trustee in a case like cheating saying something to the effect of ‘wow, you trusted me without believing – you must have really trusted me’ (Carter 2022: 3). Carter’s point here is not that either of these is the correct response, but to say that it isn’t clear whether the complaint that one isn’t trusted is ‘in any way more felicitous than praising or thanking a trustor who trusts one in absence of it’ (ibid).

Hieronymi doesn’t rely either on my reconciliation argument or explicitly endorse the distinction between trust and trusting reliance. Using both, we can provide an adequate response to Carter’s criticism. This response explains both Hieronymi’s intuition that the trustee has a legitimate complaint against the trustor, as well as Carter’s intuition that the trustee can legitimately praise the trustor for relying on them despite not believing in their trustworthiness. In cheating, Susan could have a legitimate complaint against Sam if, after years of her showing that she has been trustworthy, he still doesn’t believe that she is trustworthy.⁹ Her complaint is legitimate because Sam, even though he trustingly relies, won’t believe that she has reformed, even though she has given him overwhelming evidence to believe that she has. On the other hand, Susan could also be grateful for Sam’s trusting reliance despite his lack of belief in her trustworthiness. After all, to be treated as if one were trustworthy is still a positive thing, since it communicates that one at least hopes that one will live up to trust. It is this admirable quality that is praised in Carter’s case, though where Carter errs is in calling this thing ‘trust’. It is not trust, but trusting reliance done with the aim of letting the trustee prove themselves trustworthy to build trust back again.

Cheating provides a fourth counterexample to the reliance-based view, because it is an example of trusting reliance without trust. It also demonstrates that there can be multiple reasons for trusting reliance besides trust. While trust is, as Domenicucci and Holton note, an attitude that justifies a special form of reliance that I call trusting, it is not the only reason or attitude that can justify it. In therapeutic trust, the fact that one hopes that a trustee will be trustworthy can justify trusting reliance. There could be other justifications as well; as we saw in the literature on infidelity, sometimes a partner may recommit to a relationship after a betrayal for the sake of family, shared assets, or personal reasons, none of which relate to the trustworthiness of their unfaithful partner.

In this section, I argued against the trust-as-reliance intuition. I distinguished trust from trusting reliance. I then presented three counterexamples to the trust-as-reliance intuition which showed how it is possible to trust in cases where one does not trustingly rely. I then responded to McGeer’s objection that non-reliance views can’t account for the nature of therapeutic trust. I argued that on the non-reliance approach, to place therapeutic trust is to simply act as if one trusts. Moreover, by distinguishing trust from trusting reliance, I argued that therapeutic trust is an instance of trusting reliance without the corresponding trust. Finally, my reconciliation argument showed that this distinction makes better sense of the process of coming to trust someone again after a betrayal.

3.4 An Attitudinal Commitment Account of Trust

In this section, I propose my attitude-based, non-reliance formulation of Hawley’s commitment account of trust. I focus on a commitment, rather than motive-based approach to trust, because in a previous paper, I rejected  motives-based accounts on the grounds that, in making specific motivations constitutive of trust, they unnecessarily restrict the range of cases of trust that they can account for. I also made a distinction between ‘trusting to’ which describes three-place trust relations involving a trustor, a trustee, and an object of trust; and ‘trusting as’ which is a fourth place in the trust relation referring to the meta-relationship in which the three-place trust relation always operates (Kelsall 2022: 2). Therefore, I will incorporate this distinction into my commitment account as well.

This four-place trust relation works thus: ‘X trusts, Y to Φ, as a Ψ’ where ‘Ψ’ stands for the meta-relationship between X and Y. For example, we could say that ‘Sam trusts Susan to be faithful as his wife’ or ‘Sarah trusts Tom to look after her cat as a friend’ or ‘June trusts Joan to watch the till as her employer’ (Kelsall 2022: 10). Roughly, the argument is that it is the meta-relationships which determine whether a trustee must act out of a specific motivation to count as trustworthy. For example, in trusting as relations between friends, we might rightly expect our friends to make good on our trust out of goodwill, insofar as we take mutual goodwill to be a requirement among friends (Cocking & Kennett 1998). On the other hand, to trust a politician, we needn’t believe anything about goodwill, but could get by with trusting if we believed that they were committed to public service, for whatever reason. In other words, whether specific motivations are required for trust is not determined by the concept of trust, but by the meta-relationships that are implicit in every trust relation (Kelsall 2022: 10-11). Motives-based accounts are incompatible with trusting as relations because they require a specific motivation for every trust relation, such as goodwill.

My rejection of motives-based accounts led me to suggest Hawley’s commitment account as the most viable alternative (2022: 2). The commitment account can incorporate trusting as relations because it does not specify that trustees’ make good on commitments for specific motivations. However, Hawley’s commitment account is a reliance-based account, and I have argued against the reliance intuition in this paper. Therefore, in this section, I propose a non-reliance variant of Hawley’s account that is compatible with my notion of four-place trust, and my arguments against the reliance intuition. After introducing the account, I show how it is compatible with the idea that trust involves heightened expectations and is connected to betrayal, focusing particularly on the latter, since a central objection to my non-reliance account is that it cannot account for this intuition. Although I incorporate the trusting as-relation in my definitions, one could equally provide an attitudinal commitment account without such a feature.

Since I make the distinction between trust and trusting reliance in Sect. 2, I will provide commitment accounts for both.

Trust – X trusts Y to Φ as a Ψ if X believes that, were X to rely on Y to keep a commitment to Φ as a Ψ, that Y would keep that commitment to Φ as a Ψ.

Trusting Reliance – X trustingly relies on Y to Φ as a Ψ if X relies on Y to keep a commitment to Φ as a Ψ, and X believes that Y has such a commitment.

Except for the addition of the trusting as relation, trusting reliance is the same as Hawley’s definition of trust. This should come as no surprise, since I argued earlier that trusting reliance is equivalent to what reliance-based theorists call ‘trust’. However, to trust, one need not trustingly rely. Instead, to trust is simply to have a belief with the following counterfactual as its content: ‘if I were to rely on X to keep a commitment to Φ, then X would keep that commitment, in the context of some meta-relationship Ψ.’

In cat sitter, Sarah trusts both Tom and Anna because she believes that, were she to trustingly rely on them, they would make good on their commitment to cat sit, and the meta-relationship here is friendship.¹⁰ However, since Sarah only relies on Tom to keep a commitment and not Anna, she trustingly relies only on Tom. In regulation, organisation A trusts B and C, because it believes that B and C would keep their commitments were it able to rely on them to do so, but since it does not believe this about D, it does not trust D. Here the meta-relationship is a hierarchical, institutional relationship between regulator and regulatee. Moreover, A trustingly relies on none of the organisations since its regulative practices are incompatible with trusting reliance. This is because A relies on regulation to ensure cooperation. In cheating, Sam trustingly relies on Susan because he relies upon her to keep a commitment to fidelity, and the meta-relationship is marital love. However, he does not trust his wife because he doesn’t believe that she will make good on that commitment. He may come to trust her with time, however, if she proves that she can keep her commitment by remaining faithful.

3.5 Heightened Expectations and Betrayal

Having seen how my commitment account explains the cases we discussed in Sect. 3, we should consider whether it accounts for what philosophers take to be the defining features of trust, that is, the heightened expectations and the vulnerability to betrayal. One objection to my view is that if trust is belief, then it doesn’t involve either heightened expectations or a vulnerability to betrayal. This objection is even stronger given my account is not motive-based, as supporters of motive-based accounts argue that motives are necessary for explaining what is distinctive about the heightened expectations of trustors and the connection to betrayal (Baier 1992; Jones 1996). After all, one does not expect someone to make good on a commitment, nor is one vulnerable to betrayal, if one only believes that a trustee would make good on a commitment in some counterfactual scenario. This is a powerful objection to my commitment account insofar as vulnerability to betrayal is, as we saw in Sect. 2, a defining feature of trust. I should point out however, that the accounts of trust that define it in terms of heightened expectations and vulnerability to betrayal are reliance-based accounts, which I have argued against in this paper. Nevertheless, I can say something more to defend the non-reliance approach, by showing that the connection between heightened expectations and vulnerability to betrayal is not lost, just shifted slightly.

On my account, the heightened expectations and vulnerability to betrayal are no longer features of trust, but of trusting reliance. When X trustingly relies upon Y, she has the normative expectations that Y ought to fulfil the terms of their commitment¹¹, and if Y does not, X will justifiably feel betrayed. So, the first part of my response is to say that both heightened expectations and vulnerability to betrayal are accounted for, but via trusting reliance rather than trust. Moreover, given the value of the distinction between trust and trusting reliance, as argued for in this paper, we have good reason to accept the distinction and therefore this shift to trusting reliance.

The second part of my response is to say that although the attitude of trust does not make one directly vulnerable to betrayal, it does dispose one to make oneself vulnerable to betrayal in appropriate circumstances. This is because to trust is to believe that a trustee is worthy of trusting reliance, and through trusting reliance we make ourselves vulnerable to betrayal by relying on others to keep commitments.

An advantage of my account is that it allows us to make sense of betrayal occurring even in cases where there is no trust, but only trusting reliance. As shown in Sect. 3, we can have alternative reasons than trust to make ourselves vulnerable to betrayal, but whether trusting reliance is backed by trust or not, so long as the trustee accepts the trusting reliance, the trustor makes himself vulnerable to betrayal. Consider the following cases. First, I may not yet trust you, but want to come to trust you in the future and so I offer you a chance to earn my trust by placing my trusting reliance. This is an instance of so-called therapeutic trust. Second, I might be afraid of the consequences of not trustingly relying on you, so I trustingly rely despite my doubts. An example of this would be someone who continues a relationship out of fear of loneliness after infidelity. In either of these cases, the trustee is relied upon to keep a commitment, and if they accept that commitment and then break it, trusting reliance is still betrayed, even if the trustor didn’t trust because they didn’t believe that the commitment would be kept.

Holton (1994) considers a third kind of case, though he takes it to demonstrate that trust is reliance-based. He uses the example of so-called trust games, in which people fall back and are caught by the people below (1994: 63). In such cases, the nervous person may not trust that they will be caught, but they might fall back anyway. Holton takes this to prove that trust doesn’t require belief that the trustee will do as trusted, and that it is ultimately an act of reliance. However, such a situation can be explained as trusting reliance motivated by a desire to participate in the game, to take the risk, perhaps even to enjoy the thrill of it. Since the people behind are committed to catching you, their failure to do so is still a betrayal, regardless of what the falling person thinks about them. In the end, it is a benefit to my account and the distinction between trust and trusting reliance that we can offer a much broader characterisation of the reasons people can have for placing trusting reliance, while at the same time not sacrificing the intuitive and close tie between trust and the vulnerability to betrayal.

4 Conclusion

In this paper, I rejected the intuition that trust is a special form of reliance and developed a non-reliance variant of Hawley’s commitment trust account. I rejected the trust-as-reliance intuition by distinguishing trust, which is a belief that some trustee is worthy of trusting reliance, and trusting reliance, which is a special form of reliance. Trusting reliance involves heightened normative expectations of the trusted agent to do as trusted, as well as a justified reaction of betrayal when it is let down. Trust, on the other hand, is the belief that were we to trustingly rely on an agent to do something, that they would do that thing. After arguing for this distinction, I provided counterexamples to reliance-based trust. These counterexamples showed that one can trust without placing trusting reliance (cat sitter, enemy cat sitter, regulation), and that one can distrust while placing trusting reliance (cheating). I then argued that distinguishing trust from trusting reliance also allows us to make better sense of the reconciliatory process after betrayal. Through this reconciliation argument I showed that trust is not something that comes the moment when you trustingly rely after betrayal, but rather that, in placing trusting reliance, you give your trustee an opportunity to prove themselves trustworthy which, if successful, will hopefully restore your trust in the future. I also argued that making the distinction between trust and trusting reliance allows us to make sense of the fact that people can sometimes trustingly rely for reasons having nothing to do with the trustworthiness of their trustees.

My attitudinal commitment account builds on Hawley’s original account, my previous objection to motives-based accounts, and my rejection of reliance-based accounts in this paper. Since Hawley’s commitment account is reliance-based, I proposed an attitudinal variant of as a solution that incorporates trusting as relations. Here are my definitions of trust and trusting reliance.

Trust – X trusts Y to Φ as a Ψ if X believes that, were X to rely on Y to keep a commitment to Φ as a Ψ, that Y would keep that commitment to Φ as a Ψ.

Trusting Reliance – X trustingly relies on Y to Φ as a Ψ if X relies on Y to keep a commitment to Φ as a Ψ, and X believes that Y has such a commitment.

After introducing the account, I showed how my account can still make sense of the heightened expectations of trust and the connection to betrayal. I argued that it is ultimately trusting reliance which involves the heightened normative expectation that a trustee will meet a commitment, and that is betrayed when broken. I showed that because to trust is to believe that a trustee is fit for trusting reliance, trust is indirectly connected with heightened expectations and betrayal, since it disposes us towards trusting reliance. Thus, my non-reliance account can accommodate for the heightened expectations of trust and the connection to betrayal.