1. Socially embedded epistemic trust

Epistemic trust is a familiar enough topic within epistemology. Without trusting one’s cognitive faculties, global skepticism is assured (cf. Alston Reference Alston1986).Footnote ² Without trusting other people’s faculties, widespread but not quite global skepticism is assured. Moreover, trusting one’s own cognitive faculties may require us to give the benefit of the doubt to those of others (cf. Zagzebski Reference Zagzebski2012). Epistemic trust can be thought of as the fairly thin disposition to treat a state of some thing – a cognitive faculty, a thermometer, a testifier making noise – as a reliable enough indicator of whether a proposition is true or false. To trust someone rather than something, that is, to exercise interpersonal trust that is also epistemic, can be quite a bit richer than this thin notion.

Benjamin McCraw Reference McCraw2015 charts out a view of ET that nicely synthesizes different threads in the literature with the following account.

H places ET in S that p iff:

1. (1) H believes that p;

2. (2) H takes S to communicate that p;

3. (3) H depends upon S’s (perceived) communication for H’s belief that p; and

4. (4) H sees S as epistemically well-placed with respect to p. (425)

McCraw claims of conditions (3) and (4) of this account that they are most plausible when thought of as being or involving affective attitudes directed at the testifier as opposed to construing (3) as a merely causal or counterfactual relation or construing (4) as simply a cognitive judgment. McCraw thinks this important because it helps to capture the way in which trust in general and in its own way ET is a way of being vulnerable or open to the betrayal of another person as emphasized in the work of Annette Baier and Pamela Hieronymi (422; cf. Baier Reference Baier and Baier1995; Hieronymi Reference Hieronymi2008). Likewise, he takes an affective construal of conditions (3) and (4) to capture Karen Jones more positive perspective that trusting someone is a way of taking an “attitude of optimism” toward them because it involves a “distinctive and affectively loaded way of seeing the one trusted” (417; Jones Reference Jones1996, 4). Though Baier, Hieronymi, and Jones are talking about interpersonal trust within the context of ethics, McCraw thinks there are analogous points to be accounted for in the case of ET.

I think McCraw’s conditions are perfectly intelligible shorn of any affective attitude toward the hearer. One could with a certain clinical detachment assess that a speaker is in a better epistemic position with respect to a proposition that there is little danger of being misled by that speaker on that occasion, and hence believe the testifier. In fact, we do this on occasion. What the witness of Baier, Hieronymi, and Jones as well as McCraw’s interest in accommodating their insights underlines, however, is that for human beings, such clinical detachment is frequently not how we experience the process of trusting another person. My explanation of this fact is that trust, specifically ET, is embedded within our more than epistemic relations to other people. Outside of an epistemology classroom, testifiers are not just testifiers. They are parents, children, friends, public servants, nosy gossip, fellow citizens, or even just strangers, with a sufficient air of magnanimous approachability. Testimony and other social epistemic acts take place within a richly textured relational field.

Suppose then that one were to accept the idea that ET is frequently socially embedded. I claim that there are various affective dispositions whose epistemic relevance we often miss, and we often miss their relevance because they are socially embedded. In the next section, I will make a case that gratitude is just such a disposition. It is a way of manifesting trust in another person within the context of having received a benefit from them by representing them as worthy of a measure of positive regard in virtue of that benefit, and it is epistemically important.

Before proceeding, however, it will be helpful to give a working account of what gratitude is and to provide some initial reason to think that the kind of thing gratitude is might be relevant to the kind of thing that testimony is.

Our working account of gratitude shall be as follows. It is a reactive attitude that represents some person as having provided a benefit to oneself or to someone with whom one identifies, a benefit that of its nature or in that context reflects well on the benefactor as an agent. Representing an agent in this manner disposes one to feel positive regard toward them all other things being equal, and to be more inclined towards a range of actions including expressing thanks to them, speaking well of them to others, to doing like acts of benefaction, and to future acts of trust and reliance. This will be our working account, though I anticipate that many of the interactions between gratitude and the epistemic to be described in the next section would vary more in their details than their main thrust if we were to opt for another gloss on gratitude.

Gratitude, as I am using it here, involves representing another person as having benefited one. In epistemology, testimony refers to one person coming to believe something because another person has asserted it to be the case, thereby coming to rely on the epistemic agency of that other person for the goodness of that assertion-backed belief. Why think that extending positive regard to another person due to being benefited by them or the like has much to do with the epistemic status of a belief formed on the basis of someone else’s say so? After all, when I look at a thermometer and form a belief about the temperature, it is enough that it is reliable. One might suppose that reliance on a testifier is the same.

In order to bring into focus the relevance of gratitude to testimony and social epistemology generally, it is helpful to draw out the interpersonal dimension that accompanies testimonial interactions. One place where the relevance of the interpersonal shows up in the testimony literature is in the widespread recognition that sincerity, and not just competence, is requisite in a good testifier. Sincerity is neither an empty slot where an intent to deceive might have been, nor a simple application of one’s linguistic capacities to voicing a pre-existing belief. Sincere testimony is a good-faith attempt to use one’s abilities and background knowledge to put a hearer in a good position to share an epistemic good.

The exact nature of how one tries to benefit the recipient of testimony matters. Cognitive performance depends on effort, attention, framing, and the perceived safety, epistemic and otherwise, of the environment. Cognitive performance is not invariant, supposing only the intent to track and share the truth.Footnote ³ There is a big difference between sincerely sharing what one happens to be thinking, sincerely trying to communicate the bare minimum to which one thinks the other person is entitled, sincerely trying to represent all one’s evidence while accounting for one’s biases, sincerely trying to think carefully about something with another person while bearing their perspective in mind, and so on. Just as competence comes in different kinds and degrees, so does the ability of another person to exercise care in communication. In short, communication requires putting another person in a position to understand something, which requires more than just knowing a fact one is willing to express.

In fact, Heidi Grasswick has helpfully suggested that the sincerity condition on successful testimony be re-named the “sincerity/ care condition” (Reference Grasswick2018, 81), drawing on the care ethics tradition (cf. Held Reference Held2007). In doing so, she draws our attention to the fact that the sincerity condition, in addition to being a condition for an epistemic good, also encroaches on the domain of moral value.

She explains,

Grasswick grants that there may be one-off testimonial encounters in which a bare-bones kind of sincerity is sufficient, but over the long haul, social environments and their epistemic dimensions require more care and investment in each other, a deeper and more stable axiological orientation to one another, in order to be conducive to human flourishing. Grasswick’s discussion of sincerity occurs within the context of considering how trust can be damaged and thereby lead to epistemic injustice, but what holds in the negative direction should hold in the positive direction as well. Human flourishing requires not just the warding off of injustices but the promotion of mutual benefit. Gratitude is a reactive attitude that is uniquely qualified to affirm aspects of human relating that are conducive to human flourishing, which go beyond obligation or that manifest especially salutary ways of fulfilling our obligations of care. Consequently, how gratitude might be relevant to testimony bears investigation.

In a similar vein, in Casey Johnson’s development of care ethics in the context of epistemology, the reality of epistemic vulnerability is fundamental. She defines epistemic vulnerability as follows: “X is epistemically vulnerable to Y just in case X stands to be benefitted or harmed by Y’s actions in pursuit of X’s epistemic goals” (Reference Johnson2023, 58). Drawing on Joan Tronto’s work on political activity as care, Johnson argues that doing well by the epistemic vulnerabilities of others requires attentiveness, responsiveness, and competence that clearly is meant to encompass interpersonal competence and not just the ability to reliably express something that is true (90ff, but especially 95). Though Johnson does not mention the sincerity condition by name, she, like Grasswick, is clearly very attuned to the way in which testimony in the wild is more interpersonally demanding and affectively implicated than analytic epistemology sometimes lets on, and that interpersonally mediated reactive attitudes are important to navigating this terrain well. If this is true of reactive attitudes that seek to identify, protect, and empower the vulnerable, however, this ought to also be true of reactive attitudes that seek to identify, reinforce, and promote acts of epistemic care and the social contexts that make them possible. Gratitude is a natural candidate that bears investigation in this capacity.

While it should perhaps not be surprising that an ethics of care and its epistemic analogs would be friendly to the incorporation of more interpersonal reactive attitudes, it is noteworthy that critiques of overly naïve views of sincerity in the testimony literature that presume no such background lead to much the same place. In separate articles, Richard Moran and Stephen Wright argue that it is tempting to think that sincere testimony is nothing more than a testifier saying what they believe. They have different reasons for thinking this is an error. According to Moran Reference Moran2005, directing someone to believe something is a normative action. In doing so, one takes on responsibility for another’s beliefs, and thus an analysis of whether or not one ought to have done so or has done so well will look quite different than determining whether or not one has in fact reflected a belief accurately in language (aka sincerely), though it may certainly include that. Wright Reference Wright2014 points out that one can happen to believe what one says but relate it for a different reason (e.g. because one is trying to make a sale). By contrast, sincerity seems to require a more intimate causal connection between one’s assertion and its epistemically relevant psychological basis.Footnote ⁴

Likewise, Peter Graham Reference Graham, Michaelson and Stokke2018 critiques a Tyler Burge-inspired thought that we have an entitlement to believe speakers to be sincere due to our knowledge of how intelligibility, and thus rationality, will manifest themselves. There is more to taking someone at their word, for Graham, than assessing rationality and intelligibility. Thomas Simpson Reference Simpson2012 similarly critiques a David Lewis-inspired thought regarding how the rise of truthfulness as a convention might undergird assumptions of testifier sincerity. A convention of truthfulness does not tell one what one needs to know about a competent testifier in order for it to be wise to believe them. What these different threads of the literature have in common is the thought that what the sincerity condition should be getting at is a certain way of relating to another person, but the competence or even the intention to state out loud what one in fact believes does not by itself sufficiently specify how the testifier manages to relate well to the hearer.

Interpersonally directed reactive attitudes such as gratitude are potentially relevant to testimony because instances of testimony, whatever else they may be, are interpersonal gambits in which one person may or may not exercise a notable or appropriate degree of effort and skill on behalf of another person. In the next section, I move beyond arguing for the potential relevance of gratitude to testimony, and I make the case that gratitude is relevant in three ways.

2. Three interactions

In this section, I describe three different interactions between gratitude and the epistemic realm.

First, I claim that gratitude, even when it is not for a benefit that is epistemic in nature, has implications for how we conduct our epistemic lives downstream of that benefit, which the following case illustrates.

As regards BENEFIT OF THE DOUBT, I claim that Pam is epistemically justifiedFootnote ⁵ in according Pam’s testimony more weight than Ned’s all else being equal. Now, of course, it could be that not all is equal. If Ned is a nuclear physicist, Trish is not, and the conflicting testimony concerns nuclear physics, then Pam should discount Trish’s testimony in favor of Ned’s. If Trish and Ned are equally well-informed on the topic, so far as Pam knows, but Ned’s testimony fits decidedly better with Pam’s background knowledge than Trish’s, then, once again, that may be reason to side with Ned rather than Pam. Nonetheless, Pam should be grateful to Trish for the various kindnesses she has extended, and that gratitude should manifest itself as more easily believing Trish than Ned, not only for relational or pragmatic reasons but also for epistemic reasons.

Trish’s track record suggests that she is frequently attuned to Pam’s needs, is disposed to help her address those needs where appropriate, is willing to expend some effort to do so, does not relate to Pam in a predatory manner, and is a generally trustworthy person. Pam has a lot of evidence that Trish is invested in their relationship, and thus Pam should assume, all other things being equal, that Trish would relate to Pam’s epistemic needs similarly. Saying Pam should “assume” this is too clinical a description, however. Trish’s track record with Pam should affect how Pam represents Trish and how she feels about depending on her in this scenario. Pam should relate to her with gratitude.

Supposing then that Pam is grateful to Trish in light of their relational history, Pam’s gratitude tracks a relevant feature of Trish as a testifier and epistemic cooperator generally. Thus, feeling more inclined to extend ET to Trish in light of feeling gratitude to her would, in turn, be an epistemically beneficial disposition, not just a pragmatically, relationally, or morally beneficial one.

Furthermore, our epistemic agency is enacted over time, and, unlike thermometers, the extent to which other humans may be interested in partnering with us epistemically depends on prior interactions.Footnote ⁶ One way we can create conditions where the other person may be kindly disposed toward partnering with us in the future is by being of benefit to them, of course, but another way is by showing that we recognize and value when they are of benefit to us, such as by showing gratitude (cf. Smith et al Reference Smith, Pedersen, Forster, McCullough and Liebermann2017; DeSteno et al Reference DeSteno, Bartlett, Baumann, Williams and Dickens2010). Suppose, for instance, that the piece of testimony that Trish and Ned give demands an action that will show who Pam chose to trust on this occasion. By siding with Trish and showing affectively that Pam considers receiving this testimony a benefit, Pam reinforces her relationship with Trish, including its epistemic dimension. Even if Trish is wrong, she is more likely to continue trying to partner with Pam epistemically for Pam having shown appreciation of a past effort. Thus, even if Trish is wrong, siding with her might be diachronically rational in a specifically epistemic sense in that the epistemic goods that come with a positively valenced relationship between them outweigh the costs of potentially getting it wrong by trusting Trish on this occasion.

In fact, given these diachronic considerations and the extent to which relationships and social arrangements generally are both path-dependent and frequently embed epistemic activities and interests within more than merely epistemic bases for relating to one another, one can scale up these considerations to suggest that the web of knowledge both does and, to an extent, should map onto peoples’ non-epistemic trusted relationships and institutions, at least insofar as that trust is earned. The trustworthiness of institutions, like that of individuals, is a matter of more than mere competence. Institutions can be better or worse at employing whatever competence they possess in a way that is of benefit to people like oneself. Showing that one values and trusts these relationships and institutions constitutes a vote for their continued existence or growth. Ingratitude and taking epistemically beneficial relationships and arrangements for granted undermines them, and no matter how much competence resides in a given social network, expressing distrust in it not only presents a defeater to appropriate uptake for oneself but also undercuts, even if in a small way, the social trust that undergirds that stretch of the cognitive niche we collectively inhabit.Footnote ^7, Footnote ⁸

Having used a case to illustrate the epistemic implications of generic gratitude, I will now argue that the state of gratitude itself has an epistemic dimension with the help of a Dickens-inspired case.

The key insight here is that the transition Pip undergoes from resenting Magwitch to being grateful to him has to do with a change in how Pip represents Magwitch. This change is both epistemically motivated and epistemically committed.

One prominent theory of emotions associated with Robert Roberts is that they are concern-based construals (cf. Roberts Reference Roberts2003). In this theory, emotions are akin to a seeing as, such as when one sees a line drawing as either a duck or a rabbit depending on how one attends to it. In fact, Roberts claims of state gratitude, which he thinks of as an emotion, that each of the factors one might take to be part of an adequate conceptual analysis of gratitude – the benefit, the other having acted well by benefitting one, the other having acted from benevolence, the other having acted in excess of their duties, having been put in a “debt of grace,” etc – is built into how one construes the situation when one experiences gratitude (Roberts Reference Roberts, Emmons and McCullough2004; Reference Roberts2015).

When one thinks of or encounters a person, we do not represent all the properties we think they have equally. When we are grateful to a person, we do not merely add facts related to benefits we’ve received alongside all the others we know about the person. In paradigmatic cases, to be grateful to a person is to confer a salience or centrality to the way the other has benefited us in how one thinks of them. We anchor our understanding of who a person is and how one ought to interact with them in terms of what properties or experiences we take to be most germane to the personal meaning we find in past interactions with them and to how future interactions are likely to play out. This salience anchoring applies both when the salient past event is a negative experience, such as a harm or a wrong done, and when it is a positive experience.

When we are wronged by someone with whom we have had a close relationship, part of the struggle to forgive is the very natural tendency we have to frame our continuing interactions with the person with the wrong that has been done (Green Reference Green2024). Even if we are satisfied that the other person is penitent, that restitution has been made, and that the offense is unlikely to be committed again, we may nonetheless have a hard time viewing the other person in a way that treats these facts or our past relationship as just as or more important than the wrong that’s been done. The choice to forgive is often at its root a choice to reject a resentment-based construal of the other person in favor of a way of representing the other person that de-centers the wrong. And the difficulty of forgiveness can be a good thing. If we too easily compartmentalize a wrong, then we will not reevaluate relationships that we should or mobilize our resources to protect ourselves when we should. So, though it is important that we be able to overcome our sense of having been wronged and that many wrongs have a salience window beyond which they fade, it is also true that viewing a relationship through a wrong is an important epistemic capacity of ours. Once again, what is true in the negative direction can also be true in the positive one, however.

Pip had been disposed to represent Magwitch to himself and to others in negative terms. Even if Magwitch had acted in a kindly way on an occasion, because Pip experiences “dangerous escaped convict” as the most salient descriptor for Magwitch, Pip would be more likely to reinterpret Magwitch’s behavior as duplicitous than he would be disposed to rethink the adequacy of how he represents the man. How Pip represents the anonymous benefactor, however, follows a different track. Pip is not just affectively invested in being benefitted. He is disposed to represent this anonymous person in a way that holds benefaction as the most salient fact that pertains to this person. Without more detail about why the benefit was extended, there may be some play in how Pip represents this anonymous person, but Pip acquires a disposition to represent this person positively in virtue of the impact they have had on Pip’s life. That is, Pip’s gratitude leads him to construe the benefactor as worthy of that gratitude. For this reason, when Pip learns that Magwitch is the anonymous benefactor this causes cognitive dissonance. Pip has to reconcile two sets of facts, namely Magwitch being both an escaped con and a donor to orphans, which is perhaps not too difficult. There is no logical inconsistency here. The more demanding task is that Pip has to reconcile two different lenses for viewing a person and organize how to think about them, both of which he had gotten used to using.

When we are benefitted by someone, especially in a context that is not transactional in nature, we are invited to frame our continuing interactions with that person through the window of their kindly intentions toward us.Footnote ⁹ All manner of psychological resistance might stand in the way of doing that, and depending on how it is calibrated, it might be a very good thing to have a measure of psychological resistance to too easily revising one’s view of another person due to being benefitted by them on an occasion. To choose to be grateful, though, is to choose to side with or voluntarily adopt the fact that someone has been good to us as a lens through which to relate to them and to think about one’s relationship with them. The cognitive content of gratitude, thus, does not just include the fact of one’s benefaction. It represents the giver as worthy of certain kinds of affectively positive responses, and it represents relationship with the giver as to one extent or another worth investing in.Footnote ¹⁰ And, of course, these facts about the world that are so represented can be right or wrong.

If one’s gratitude is apt, then one has some reason to suppose that the other person is, to one extent or another, someone to trust, someone who would be good to cooperate with in the future. And as our BENEFIT OF THE DOUBT case illustrated, that makes them epistemically relevant to our lives just in virtue of that fact. Choosing to be grateful to a person, even if it is for a benefit that is not obviously epistemic, is nonetheless, in part, an epistemic act with epistemic implications. But, of course, the benefit in question could be of a distinctly epistemic nature as well.

Thus, finally, let us look at gratitude for specifically epistemic benefits.

There are plenty of non-epistemic goods that are important to Connor’s work with his two trainees. We can say, for instance, that Connor is compensated monetarily for his time training them, that the business will do better if its trainees are mentored well, that Brett and Lisa will make more money because of Connor’s efforts, etc. Connor’s mentoring relationship with Lisa is distinct, however, because Lisa does not treat Connor’s epistemic help as something to be taken for granted within the context of achieving these other non-epistemic goods. By showing gratitude towards the specifically epistemic benefits that come from Connor’s mentoring, Lisa draws Connor’s attention to this dimension of their relationship, which puts him in a better position to be intentional about it. He no longer takes this part of mentoring for granted because her gratitude draws it to his attention,Footnote ¹¹ and in fact, because it is a positive experience to have someone demonstrate gratitude towards you, Connor acquires an incentive to focus on the specifically epistemic dimensions of the mentor-trainee relationship. Moreover, because gratitude depicts one as worthy of that gratitude, Connor acquires an incentive not just to share what he knows but to enact his own epistemic agency in a way that accords with Lisa’s gratitude, to live up to her high regard.Footnote ¹²

Likewise, from Lisa’s side, feeling and expressing gratitude for a specifically epistemic good also serves to highlight the relevance of such goods. It makes her less likely to overlook or merely instrumentalize epistemic goods, in particular within this context.Footnote ¹³ Because she places more value on her relationship with Connor than Brett does, she is more invested in it, allocates more attention to it, and, as a result, is better positioned to benefit epistemically from the relationship even when Connor treats the two of them the same. Feeling gratitude for epistemic mentoring makes her more likely to think about the epistemic dimensions of her other work relationships or to consider in what contexts she might similarly be of benefit to other people epistemically. To depict another person as worthy of gratitude can, in turn, provide one a reason to want to become the sort of person who is worthy in similar respects.Footnote ¹⁴

Finally, knowledge in this example is a shareable good.Footnote ¹⁵ Connor does not have less of it because Lisa has more of it. Though there are some non-shareable goods in the vicinity, such as having exclusive access to knowledge, treating the shareable goods that are distributed in a relationship as highly valuable and deeply appreciated changes the context in which people think about exchanging goods that are not shareable. If the most salient goods to a relationship are not shareable, then risk management and fairness monitoring become of paramount importance.

In Connor’s relationship with Brett, for instance, he is obligated to put Brett in a position to help the business, but he should think twice about the risk of sharing too much with Brett, since, for all he knows, what Brett values are goods that could be pursued at his expense in the right circumstances (such as Brett’s being promoted over Connor). By contrast, although risk management and fairness monitoring surely would not be absent, Lisa’s valuing of goods in her relationship with Connor which are shareable, epistemic goods being an excellent example of such a good, allows the worrisome features of non-shareable goods to have a relational counterweight which encourages more trust and collaboration. Moreover, this increased trust is not unreasonable. Though conflicts can certainly emerge, one should, in fact, expect a person to be less inclined to be thoughtless, selfish, or undermining toward someone else to whom they relate with genuine gratitude.

Being attuned to being benefited by others epistemically and demonstrating that gratitude can also be generative of value. The fact that Lisa shows gratitude for being epistemically mentored provides a reason for Connor to want to continue investing in Lisa epistemically, but this need not be asymmetric. It also incentivizes trying out ideas and conducting new inquiries together. If there is at least some intrapersonal and interpersonal reward from engaging in the life of the mind together, then one is more likely to engage in it even when it is not strictly required or its pragmatic implications are not demonstrable up front when one has a collaborator with whom one share good affective rapport. This, in turn, can be generative of new epistemic goods and generative of new non-epistemic goods based on the application of those new epistemic goods. Valuing the epistemic dimension of a relationship does not just spread existing shareable goods further but is generative of value itself.

As with the epistemic relevance of generic gratitude, the importance of gratitude for specifically epistemic goods scales up. As the distrust of public health institutions during the Covid pandemic (cf. McCaughlin et al. Reference McCaughlin, Mehwhirter and Sanders2021) and the continuing ambivalence towards climate change experts (cf. Hmielowski et al. Reference Hmielowski, Feldman, Myers, Leiserowitz and Maibach2013) amply illustrates, distrust in social infrastructure that serves important epistemic functions in society is not necessarily a result of not understanding how science works. That is, the problem is frequently not doubting the competence of scientists or organizations like the Center for Disease Control and Prevention (CDC). The primary problem rather is often a suspicion of the motives of public health experts, climate scientists, and related organizations and specifically their politicization. In the terms of the testimony literature, the problem is not competence but the perceived sincerity of these experts. Climate deniers and Covid skeptics share an affective orientation, a lack of the sort of positive regard that would confer enough resiliency on our dependence on scientific experts and relevant institutions as epistemic sources to not be easily swayed by misinformation, conspiracy theories, and the occasional public relations gaff. In brief, one might fruitfully diagnose these social trust problems as in no small part failures of gratitude for past and present benefits rendered due to a kind of ideological capture of public perception.

In sum, in this section, I have argued for the epistemic relevance of gratitude in three ways, by pointing out its relationship to ET and the sincerity condition on testimony, by showing ways in which gratitude itself is epistemically evaluable and subject to epistemic agency, and by describing the importance of gratitude for specifically epistemic goods. I have not claimed that gratitude is essential condition for knowledge, justification, or any other epistemic good. Yet, its relevance to epistemology is more fundamental, or so I claim, than just any way of having positive affective regard for another. The reason is this. Gratitude is our most basic reactive attitude for relating positively towards another person by virtue of their relational track record with us. Therefore, it is affective ground zero for relating to other people epistemically by virtue of their track record with us. Moreover, because our epistemic lives are enmeshed with our lives more generally, as I brought out in section I, the ways in which gratitude or failures thereof build our relationships and social institutions play an important role in establishing the social contexts in which we pursue knowledge generally.

3. Gratitude gone bad

As will no doubt not be a surprise, gratitude can misfire or be exploited. In this final section, I want to diagnose what goes wrong from an epistemic perspective when gratitude goes bad. My contention is that the ways gratitude can go wrong, far from being evidence that gratitude is not epistemically significant, are reasons to think that, because gratitude is epistemically important, it is also important that gratitude be well-calibrated. Moreover, thinking about the ways gratitude can go bad alongside the ways that gratitude is important to our epistemic lives helps to bring into focus what it would look like to promote epistemic gratitude in a way conducive to the goods that come with it.

It is not hard to imagine each of the three scenarios from the previous section taking a turn in a dubious direction. Suppose it turns out Trish is part of a cult and starts trying to actively recruit Pam. Suppose Magwitch tries to use Pip’s indebtedness to him to initiate Pip into a life of crime. What if Connor gets creepy with Lisa? Clearly, the benefactor-beneficiary relationship can be turned in malign directions, including in its epistemic dimensions. Thus, one might think that gratitude is risky and that it is risky in no small part because of the way it lends itself to enduring ties with other folks in a way that might seem to be suspiciously independent of epistemic value.

I want to highlight three different features of the kinds of cases where it looks like gratitude yields bad results, epistemic and otherwise.Footnote ¹⁶ To illustrate these points, we’ll use the * to designate the nearest possible worlds where our good cases of gratitude have run amok.

First, gratitude represents the benefactor as worthy of positive regard. Two ways in which this representation can go wrong are that it can represent an unworthy benefactor as worthy or it can represent a benefactor who committed an isolated good action as having enduring traits they do not have thereby overgeneralizing the positive regard that is in fact due to them. For instance, take the scenario where Pam* represents Trish* as an exemplary, caring friend and neighbor, and Trish* starts trying to recruit Pam* into a cult. Trish*’s past actions toward Pam* (e.g. watering her plants; and being a supportive listener during difficult times) are worthy of an affectively relevant response, but, of course, exactly how probative one takes these occasions to be can vary. Suppose, for instance, that Pam* experiences significant betrayals in her other relationships such that, against that background, Trish*’s positive qualities take on a certain luminosity. Pam* acquires a disposition to trust Trish* in a way that goes beyond her evidence because of the way Trish* contrasts with these persons who have betrayed Pam*’s trust. In that case, Pam*’s gratitude makes her vulnerable to not taking a nuanced enough view of Trish* and thus not identifying and resisting Trish*’s cult involvement the way she would if she did not feel gratitude towards Trish*.

Second, in cases of gratitude gone wrong, often it is not only the benefactor that is misrepresented in some critical respect but the benefit. One critical way in which this can occur is when the benefactor initiates a transaction under the guise of giving a gift. Take, for example, a scenario in which Magwitch* tries to initiate Pip* into a life of crime, saying that Pip* owes Magwitch* certain favors by virtue of everything he has done for Pip* over the years. If Magwitch* had approached Pip* from the beginning with a proposed transaction, money for Pip*’s education in exchange for breaking a few thumbs and collecting a few debts, then, for all that might be wrong with such a scenario, it would not feature a malign form of gratitude per se. It would just be a bad proposed transaction. The distinctive contribution of gratitude to the manipulation of Pip* is that Magwitch* in fact hides the transactional nature of how he intends to relate to Pip* behind the laudatory moral logic of benefaction.Footnote ¹⁷ Then, once Pip* already experiences himself as in a position where gratitude is appropriate in virtue of the lavish gift he has received and thus Pip* already feels he should demonstrate his gratitude in action if given the opportunity, Magwitch* changes the nature of the gift.

Relatedly but more subtly, changing a gift into a transaction devalues the recipient. To give someone a good gift is to invite them to view themselves as worthwhile in some sense (cf. Berger Reference Berger1975; Card Reference Card1988, 117). Whatever a “debt of gratitude” may be, a gift is a way of benefitting another person without any guarantee that one will be similarly benefitted in turn. Since the conferring of benefit cannot be made rational by an expected return on investment in kind, its rationale must lie elsewhere. A gift implies that there is something about the recipient that makes being good to them apt. Changing a gift into a transaction, as it were, cancels that implicature, substituting a mode of relating that demands that benefitting the recipient be made worthwhile by holding up one’s end of a bargain. Thus, in addition to Magwitch*’s manipulation of Pip* which already disrespects Pip* as an autonomous agent, the form that manipulation takes further devalues Pip* by using a way of representing Pip* as intrinsically worthwhile to relate to him in a way that ignores or denies that worth.

Third, not only can one misrepresent the character of the benefactor or the nature of the gift but one can misrepresent the kind of relationship that gratitude should move one to invest in. Suppose that the all too familiar scenario plays out such that Connor*, through the course of developing a positive mentor-mentee relationship with Lisa*, begins overstepping the bounds of his role. Suppose Lisa* initially feels cognitive dissonance at this development, dissonance she might not feel if she were not genuinely grateful for their working relationship, at least to this point. She might otherwise simply be creeped out and on guard, if not immediately on the phone with human resources. Lisa*’s gratitude represents her relationship with Connor* as a good thing and worth investing in, but it runs the risk of making her more vulnerable than she would otherwise be to Connor*’s inappropriate advances. The distinctive contribution of gratitude to this situation is that Connor*’s (pre-creepy) generous mentoring does in fact merit updating Lisa*’s picture of the world in a way that values that relationship. Moreover, properly valuing someone qua one role can in fact be good evidence for how one should value a person in general or in a different role. Nonetheless, the implications of gratitude for how one ought to relate to a person going forward can be over-extended or unclear, which can, in turn, be taken advantage of, as is the case with Connor’s creepy doppelganger Connor*.

These scenarios all involve epistemic problems, but they are far from inevitable ones. I see no reason to think of them as showing that gratitude is epistemically bad, risky, or too dangerous as such. I conclude, rather, that gratitude’s epistemic benefits hinge on engaging in gratitude well in an environment that is friendly to its operation. Gratitude needs to be apt, skillful, well-calibrated, and reliable. This is unremarkable insofar as the same could be said about cognitive faculties like memory or perception or for dispositions such as looking for counterevidence or a tendency to double-check one’s inferences. It may well be, though, that gratitude functions best when paired with other qualities such as appropriate self-respect and an awareness of the human susceptibility to abuse power. Likewise, it may function best when enacted in contexts where one is not overly dependent on any one individual and where taking advantage of power asymmetries is appropriately disincentivized.

The risks of gratitude should also not be assessed without considering the alternative. Suppose we were to successfully insulate our epistemic lives from gratitude. Insofar as gratitude is the way we come to represent that someone has benefited us due to good properties they possess and relate to them accordingly, numbing ourselves to gratitude impairs our ability to note, care about, and act on this state of affairs. Likewise, showing gratitude is an important means at our disposal to incentivize people to benefit others, affirm relationships in which we benefit, and indicate our investment in institutions that do good in our lives. We should not take a disinterested stance toward such things and to do so is to be careless toward the underpinnings of flourishing for a social creature. One of the goods that our flourishing as social creatures is dependent on is the generation, dissemination, and defense of good information through the epistemic generosity of others, through the epistemic dimension of our enduring relationships, and through social institutions and cultural scaffolding that produces public knowledge and form the epistemic basis for common action. Gratitude, as the core prosocial reactive attitude concerned with registering and responding to being benefitted by others, is essential to this project.

By way of conclusion, let us revisit the relationship between gratitude and trust. To be grateful to a person is to represent them as having benefited one in some way that reflects well on them. Thus, to be grateful to a person is to represent them in a way that gives one a reason to trust them, albeit a defeasible reason. In fact, one might say that gratitude frequently just is the interpersonal attitude that expresses our trust in those who have been good to us. Trust in persons is interpersonal and affectively implicated, and because the person we trust must be motivated to continue to benefit others, demonstrating one’s gratitude to others is an important task in maintaining relations of trust. This is an important asymmetry when it comes to ET in persons and non-persons. If one trusts another person but fails to be grateful to them or to express it, one’s ingratitude is a failure to care for the epistemic niche, the web of distributed knowledge, that one benefits from.Footnote ¹⁸