1. Introduction
As limited beings, we rely on others to acquire knowledge. Sometimes, these others are more knowledgeable than we are, especially in highly specialized domains. Often, they are our peers who happen to be in a better position to have knowledge about a specific issue. In the name of epistemic economy, it is wise to rely on them rather than checking everything ourselves. The question is when and under what conditions we are entitled to do so.
Over the past five decades, one of the most significant topics within social epistemology has been whether testimony is a basic source of prima facie justification. In other words, are we justified in believing p based on the testifier’s testimony only if we have positive evidence of their trustworthiness (E. Fricker Reference Fricker, Lackey and Sosa2006), or are we prima facie entitled to believe p based on their testimony in the absence of counterevidence (Burge Reference Burge1993; Graham Reference Graham, Haddock, Millar and Pritchard2010; Simion Reference Simion2021)?
In this paper, without questioning the results of the classical approach in the epistemology of testimony, I propose to consider the testimony from the side of our daily practice. I inquire what separates the default epistemic entitlement and our actual entitlements as empirical agents:Footnote 1 the evidence we have at our disposal on a daily basis when we have to decide whether to trust S or to demand additional justification. In fact, we are rarely in a default situation where our testifier is completely anonymous to us.Footnote 2 Even if we are wrong about them, due to the stereotypes and prejudices (Begby Reference Begby2021; M. Fricker Reference Fricker2007), we already ascribe a certain epistemic position to our testifiers in advance and tune our rational epistemic trust accordingly. I focus on a specific type of evidence we use daily, epistemic reputation.
Epistemological work on reputation has gained traction in recent years, particularly through the work of Origgi (Branch et al. Reference Branch, Origgi and Morisseau2022; Origgi Reference Origgi2012, Reference Origgi, Giardini and Wittek2019a, Reference Origgi2022; Origgi et al. Reference Origgi, Holmes and Arikha2018). Despite this significant research, reputation has remained largely overlooked in debates about the nature of testimonial knowledge and the epistemic entitlements of receivers of testimony, in particular. My goal is to propose a novel account of epistemic reputation and determine its evidential weight. I define epistemic reputation as the social standing of agents, groups, or institutions within a given community, grounded in their perceived epistemic profiles. Using a functionalist approach, I argue that epistemic reputation is pervasive in our societies because it helps us navigate social epistemic networks. By this I mean that reputation provides crucial cues about the epistemic positions of other agents, groups, and institutions, both in terms of how strongly warranted their beliefs are and in terms of the quality of epistemic agents, groups, and institutions, which is reflected by their faculties, virtues, and vices.
The plan of the paper is as follows. In Section 2, I delve into Origgi’s account of reputation and establish its relevance to the epistemology of testimony. Section 3 introduces my functionalist approach to reputation and presents its evidential position. Section 4 then explores how reputation is constructed based on its identified function. Section 5 examines instances of misleading reputation, while Section 6 delves into the epistemic responsibilities associated with reputation, particularly for the society and experts.
2. Why reputation matters?
The epistemology of reputation has primarily been explored by Origgi (Branch et al. Reference Branch, Origgi and Morisseau2022; Origgi, Reference Origgi2012, Reference Origgi, Giardini and Wittek2019a, Reference Origgi2022; Origgi et al. Reference Origgi, Holmes and Arikha2018). Her project focuses on analyzing how reputation emerges as a social source of information and potentially knowledge for us. Origgi begins by noting that we don’t directly perceive people or items as such. Instead, we always consider them in the context of a specific structure to which they belong (Origgi Reference Origgi2012: 409). This structure is highly hierarchical, and the position an individual holds within it is important first for our actions and then for our cognition. When interacting with others or selecting items, we rely on the hierarchical structure to determine their suitability for our practical needs. Similarly, when seeking expertise, we evaluate the individuals occupying relevant positions in the social hierarchy to establish trust in them. In all of this, we rely on a specific type of social information – epistemic reputation. Origgi defines it in the following way: “Reputation is a special kind of social information: it is social information about the value of people, systems and processes that release information.” (Origgi Reference Origgi2012: 401)
Origgi’s research, though acknowledged as a part of social epistemology (see Origgi Reference Origgi, Fricker, Graham, Henderson and Pedersen2019b), remains marginal to normative social epistemology, particularly to the epistemology of testimony. One reason for this may be that Origgi draws inspiration primarily from studies in economics, social psychology, and anthropology, and she employs frameworks and terminology borrowed from these disciplines. Consequently, she speaks in terms of information management rather than beliefs, knowledge, and so on, which hinders her significant contribution from fitting seamlessly into the framework more natural for normative epistemologists.Footnote 3
The marginalization of reputation in the epistemology of testimony represents a significant loss. As I will argue later in this paper, reputation is not a peripheral aspect of social life that might incidentally interest epistemologists – whether in terms of how we acquire information about it or how we treat evidence based on it. Rather, the epistemology of testimony ought to attend specifically to a distinctive kind of reputation, namely the reputation of testifiers themselves. In many cases, when evaluating a testifier’s credibility, we lack direct evidence grounded in personal experience – be it from a shared epistemic history or from firsthand knowledge of the interlocutor’s epistemic competence. In such situations, we frequently rely – inevitably and often unreflectively – on the testifier’s reputation as a shared social epistemic resource. Given this, it is crucial that the epistemology of testimony engages seriously with the normative status of this specific form of evidence: epistemic reputation.
To see why it matters, consider the problem of expert status. Classically, the epistemology of expertise poses two questions. The first is objective and concerns the criteria for expertise (Croce Reference Croce2019; Croce and Baghramian Reference Croce and Baghramian2024; Farina et al. Reference Farina, Lavazza and Pritchard2024; Goldman Reference Goldman2018). For instance, Goldman defines expertise as follows:
S is an expert in domain D if and only if S has the capacity to help others (especially laypersons) solve a variety of problems in D or execute an assortment of tasks in D which the latter would not be able to solve or execute on their own. S can provide such help by imparting to the layperson (or other client) his/her distinctive knowledge or skills. (Goldman Reference Goldman2018: 4)
Goldman delves into the nature of expertise, exploring whether it is a reputational or reputation-independent phenomenon. In other words, he questions whether public recognition of S as an expert is necessary for S to be an expert. He rejects the reputational understanding of expertise because he is interested in the objective status of an expert. For effective assistance, it suffices for an expert to possess the relevant abilities or knowledge. They do not need to be recognized as such.
However, challenges persist. An unrecognized expert S may possess the necessary expertise but may not be perceived as such by an addressee A, especially in high-stakes situations. From A’s perspective, trust in S necessitates recognizing them as an expert. This raises the second question of epistemology of expertise: how to recognize an expert? Goldman (Reference Goldman2001) addresses this problem in contexts where two competing candidates claim expert authority. He identifies five potential sources of evidence for determining expertise: (1) the arguments presented by the contending experts, (2) agreement from additional putative experts, (3) appraisals from meta-experts, (4) evidence concerning the candidates’ interests and potential biases, and (5) their respective track records.Footnote 4 Notably, Goldman approaches this issue from the perspective of an individual agent: while they may draw on these various criteria to assess whether S is indeed an expert, they are ultimately responsible for making their own judgment.
Epistemic reputation, I suggest, may serve to bridge the gap between perplexed novices and actual experts. Departing from Goldman’s still-individualistic framework, the novice need not personally gather and assess all relevant evidence. Instead, they may rely on epistemic reputation, which often implicitly conveys precisely the sorts of evidence Goldman enumerates. Two components of Goldman’s framework appear particularly aligned with reputation: first, appraisals by meta-experts (including credentials and institutional recognition); second, evidence of an expert’s prior track record.Footnote 5 This alignment supports the idea that epistemic reputation can function as a distinctive form of social evidence – one that significantly reduces the epistemic labor an agent must undertake in acquiring testimonial knowledge.
Of course, this proposal cannot be accepted uncritically. Shifting the focus from actual expertise to social standing introduces risks. On the one hand, epistemic reputation may overestimate a putative expert’s authority, reinforcing positive stereotypes that stem from entrenched social and political privileges. On the other hand, exclusive reliance on socially acknowledged experts may cause us to overlook genuine experts, particularly those who belong to historically oppressed groups (a case of systemic epistemic downgrading), or individuals who, for contingent reasons, have not yet been publicly recognized – such as anonymous contributors or newcomers to a field.
These concerns will be addressed in detail later in the paper. For now, I note that I do not claim reputation to be infallible, or even highly reliable, evidence of epistemic authority. Rather, I will be content to defend the view that epistemic reputation can provide sufficiently reliable evidence – evidence that, while fallible, should nonetheless form part of the epistemic agent’s evidential base when evaluating (expert) testimony. In most cases, reputation will be only one among several available pieces of evidence. Yet, in some situations – particularly those characterized by limited access to information – it may be the only evidence available. Even if it might falsely attribute expertise or miss some experts, I contend that reputation can serve as a reasonable, defeasible basis for epistemic trust. In either scenario, its chief epistemic value lies in its ability to externalize and distribute the task of expert recognition: society, through various epistemic practices and institutions, has already undertaken that work on behalf of the individual agent.
3. Function of reputation
Origgi’s work in the epistemology of reputation offers a valuable phenomenology of how reputation serves as a social cue that can influence our actions and cognition. I agree with her that it is an empirical fact that we are guided by reputation in both our actions and cognition. Additionally, reputation often appears to be an apt and effective tool in practical and theoretical deliberation. However, this position requires justification for why trusting reputation can be an epistemically valid way of forming beliefs. In this section, I propose a theoretically economical step from Origgi’s phenomenology to a normatively grounded account of epistemic reputation, employing a functionalist approach. In this and next section, I present a general argument and account for rational dependence on reputation. In Sections 5 and 6, I further explore the conditions that make trusting epistemic reputation more or less justified.
Our epistemic social environment is diverse, which brings both advantages and disadvantages. We differ in various aspects. One way to illustrate this is through knowledge. Traditionally, epistemic evaluation has focused on the body of beliefs. However, we can also examine epistemic agents themselves and recognize their intellectual faculties and virtues. Epistemic diversity offers numerous advantages. Specialization in specific areas allows others to acquire expertise in diverse domains. On the other hand, since there are no default epistemic agents, there is no one universally right epistemic position we should take toward fellow testifiers. People are worth our trust to varying degrees in different domains and under different contextual constraints. The same level of default epistemic trust would be too high in some cases and too low in others. We would epistemically fail by unjustifiably placing too much trust in some agents (losing epistemic warrants, potentially knowledge, and even truth) and by ignoring the valuable testimony of epistemically better-situated agents (easily missing new knowledge). Generally, we can be optimistic that, from an objective perspective, our society possesses sufficient knowledge and expertise, despite its limitations and shortcomings in certain areas. The problem is that without a proper recognition of who actually is an expert and who is not, it becomes very difficult to effectively utilize our epistemic social resources.
A useful comparison can be drawn with a metro system. What makes a metro system effective? First, we consider its plan. Stations should be strategically located in the city, the routes of the lines should prioritize major migratory directions, and the transfers between lines should be easy for commuters. Additionally, the train schedule should be punctual and frequent. Comfort is also crucial, both for the stations and the trains. We expect them to be ergonomic, accessible, well-lit, ventilated, and clean. Lastly, and most importantly, a fair fare plan is essential to promote public transport and reduce traffic pollution, combat topographic, economic, and social exclusion.
Imagine if this otherwise perfect system had faulty or opaque marking: people would wander aimlessly in corridors, searching for the right platform or exit, corridors would get clogged, journey times would lengthen, and in the worst-case scenario, passengers might not reach their destination. Failure to meet the condition of clarity could undermine all the advantages we’ve gained so far. As it turns out, clarity isn’t just an esthetic issue; it can have serious consequences for the functioning of complex systems.
The same is true of epistemic social networks. Just like a good metro system needs a clear marking system, we need not only good epistemic agents but also some social cues about whom and in what to trust epistemically. Epistemology has traditionally focused on the first set of questions: the actual quality of epistemic agents, beliefs, processes, and so on. However, as the metro analogy suggests, given the social distribution of epistemic competences and achievements, the second set of questions, concerning the proper recognition of these qualities, is equally important.Footnote 6
Thus, we need a tool to navigate the epistemic social jungle. Ideally, it should be easily accessible to individuals without specialized expertise or advanced epistemic faculties. As Goldman (Reference Goldman2001) demonstrated, it can be challenging for a novice to assess the relative expertise of different testifiers. The only reasonably safe and accessible criterion for them is the track record, but this requires some epistemic work. If there were a tool that could do this work for novices, we could significantly reduce epistemic effort at both the individual and community levels, allowing us to better utilize our epistemic resources.
One way to address this need is through epistemic reputation. I propose that the function of epistemic reputation in a community is to signal the epistemic standing of different agents within the epistemic social environment. Inspired by Graham (Reference Graham and L.R.G.2023),Footnote 7 I will first explore what it means for epistemic reputation to have signaling as its proper function and argue how this function translates into epistemic entitlements.
Graham proposes the following organizational account of functions:
A function of trait T in S is F iff:
-
(1) S is a differentiated self-organizing system;
-
(2) T does F in S;
-
(3) By doing F in S in the past, T contributed to the self-maintenance of S;
-
(4) T thereby persists now in S (at least in part) because (3). (Graham Reference Graham and L.R.G.2023: 267)
Let’s examine how the proposed function of epistemic reputation aligns with Graham’s criteria.
First, societies, and communities in particular,Footnote 8 are differentiated self-organizing systems. This premise requires no further argument, and Graham (Reference Graham and L.R.G.2023: 269) himself acknowledges that this is one of the advantages of his new account of functionalism – its ability to apply to functions operating within society.
Second, reputation serves as a signal of the status of different epistemic agents within a community. For instance, when I want to inquire about Catalan Romanesque reliquaries, I prefer Thomas over Martin. This is because my social circles share the belief that Thomas has a doctorate in art history and has demonstrated expertise in answering similar questions in the past. In contrast, I lack the reliable means to assess Martin’s expertise, and as a layperson, I am not competent to evaluate either Thomas’ or Marcin’s expertise on my own. In such cases, epistemic reputation plays a valuable role in epistemic economics. It’s not limited to experts, though; it can also help rationally acquire testimonial knowledge from epistemic peers and even epistemically weaker agents. For example, if I want to inquire about the condition of my hospitalized friend, I should accept the testimony of Kate, who has a reputation for prudence, rather than that of Matt, who is known for exaggeration.
Of course, one’s empirical reputation may be undeserved and, therefore, unreliable in assessing credentials to (epistemic) agents. More on this will be discussed in Section 5. For now, two things can be anticipated. First, epistemic reputation is one kind of evidence about the epistemic standing of agents, groups, and institutions. As such, it falls within the broader category of second-order evidence (meaning evidence about the reliability of first-order evidence, such as evidence about testifiers). It can be defeated by other elements of second-order evidence and other types of defeaters. Second, in Graham’s terms, epistemic reputation can be understood as “warrant-grade-one” that is a warrant that a belief enjoys if and only if it is “based on a (normally acquired) normally functioning belief-forming process,” and if this process has “forming true beliefs as a function” and “reliably forms true beliefs in normal circumstances through normal functioning” (Graham Reference Graham and L.R.G.2023: 256–257). This means that if there are no defeaters available for the agent, reputation-based testimonial belief is justified. However, in abnormal circumstances,Footnote 9 reputation will not be a reliable way of assessing agents’ epistemic standings. In Section 5, where I develop this further, I explain how we can still argue that epistemic reputation breeds epistemic entitlements, even though they may be unreliable in individual cases.
Returning to Graham’s definition of function, third, by signaling well the epistemic positions of different agents in the community in the past, epistemic reputation contributed to the self-maintenance of the community. The primary benefit is simply acquiring genuine knowledge by accepting its transmission from well-credited agents. When accepting an assertion based on an individual’s reputation, the addressee not only accepts true beliefs that have been properly acquired by the testifier but also meets the knowledge criteria. This is true in both internalist (reputation acting as relevant evidence for accepting an asserted belief) and externalist terms (believing from a reliable source). Moreover, the society as a whole gains in efficiency. Agents don’t need to acquire subject matter knowledge independently or second-order evidence about the credibility of other agents. It would require significant time and effort to independently verify an agent’s reliability, competence, truthfulness, and bias. This evidence, in contrast, is originally acquired by the communityFootnote 10 and is accessible to members of social epistemic networks. Additionally, reputation facilitates the acquisition of highly specialized knowledge, as most of the public lacks not only competence in a specific field but also the ability to judge experts in that field. Reputation (often expressed through various scales, accreditation systems, and so on, as mentioned by Origgi) effectively addresses the novice-expert problem (see Brennan Reference Brennan2020; Hinton Reference Hinton2020).
Fourth, reputation persists now in the epistemic community (at least in part) because by signaling well the epistemic positions of different agents in the community in the past, it contributed to the self-maintenance of the community. Imagine a dystopian scenario where reputation devolves into mere false pretenses and systematically misrepresents one’s epistemic status. This could lead to both undeserved high epistemic status (more on that in Section 5) and epistemic injustice (more on that in Section 6). If nothing were at stake, we might continue to believe in this illusion. However, since our beliefs have practical implications, if relying on reputation systematically misled us and resulted in practical failures, we wouldn’t trust it anymore and wouldn’t be guided by it in the community. Instead, we would likely revert to a more reliable (though less economical) method of determining ourselves the epistemic positions of other agents, groups, and institutions.Footnote 11
To summarize this section, I propose the following function for epistemic reputation:
The Function of Reputation is signaling the epistemic standing of different agents in the epistemic community.
The advantage of this approach is that if we accept proper functionalism, we get the etiological normative confirmation of the epistemic position of reputation. When “a belief-forming process has the forming of true beliefs reliably as a function, normal functioning encodes reliably getting things right” (Graham Reference Graham and L.R.G.2023: 268). Consequently, it provides justification to the beliefs formed in that process (Simion Reference Simion2019), although it may not always be sufficient for knowledge, especially in abnormal environments. Epistemic reputation is our tool for gaining second-order evidence about the epistemic standing of individuals, groups, and institutions. As long as the process that produces reputation functions reliably enough,Footnote 12 reputation remains a valuable asset in community, and beliefs formed on the basis of reliably produced reputation are epistemically justified.
4. On the nature of epistemic reputation
Having established the function of reputation, I can proceed to the next stage of analysis. According to the principle “form follows function,” let’s delve deeper into what constitutes an epistemic reputation. I propose a model of epistemic reputation as epistemic profiling of agents. (This is, of course, a specific model, and its narrative should be viewed as illustrative.Footnote 13) I will proceed as follows: Suppose a tool, X, that performs a function, F. What would I expect X to be in order to excel at fulfilling F? In other words, what additional requirements does F impose on X? By following this line of reasoning, we can transition from the function of reputation to the elements that make up the epistemically normatively significant concept of reputation.
I propose to start from the notion of epistemic profiles of agents.Footnote 14 By epistemic profiles, I mean the general characterization of how we perceive other epistemic agents. The purpose of such characterization is to gather information to determine an individual’s position in the epistemic social network and, consequently, the reasonable epistemic trust we should place in them, in a given field and practical context.
What constitutes epistemic profiles of agents? First, we can examine our fellow agents in terms of their body of beliefs. This body of beliefs can be categorized at least at four levels: content, epistemic status (truth, justification, and knowledge?), coherence, and practical value. We can and should stratify epistemic agents based on the quality of their beliefs. Such profiles, while confronted with our epistemic needs, prompt us to choose whom to trust and to what degree.
However, it’s crucial to recognize that this is merely one of the ways in which our social communities differ epistemically. Therefore, we should also focus on the epistemic faculties, virtues, and vices of agents. It’s worth revisiting Craig’s story of identifying reliable informants (Craig Reference Craig1990; see also Reynolds Reference Reynolds2017). When seeking good informants, we should consider not only the knowledge S possesses but also their epistemic profile: the quality of their epistemic faculties and responsibilist virtues that make them a trustworthy testifier. In many cases, we may initially be uncertain whether S has knowledge about p. However, knowing what kind of epistemic agent S is provides reasons to believe that we might (and should) defer to them if we wanted to know whether p.
Let’s observe that this approach helps us understand the appropriate social role of many epistemic responsibilist virtues whose value may not directly lead to producing epistemic goods. In a conversational situation where we seek knowledge, two conditions must be met for us to acquire that knowledge through testimony. First, S must acquire the right belief. Second, S must ensure that the proper norms of assertion are met to facilitate the transmission or generation of epistemic status of beliefs (Kelp and Simion Reference Kelp and Simion2021). Neither condition is automatically fulfilled. Both parties can benefit from exercising relevant epistemic virtues. And these virtues can serve as reliable markers for us to justifiably assess S’s position to rely on them concerning p.
On the one hand, virtues like open-mindedness, inquisitiveness, and epistemic courage aid S in acquiring knowledge. Conversely, recognizing that S embodies these virtues gives us the grounds to rationalize trust in them. On the other hand, mere knowledge acquisition isn’t enough for S to be a reliable testifier. Every good testifier must possess certain traits of character and abilities that (1) make them trustworthy, (2) ensure the accurate transmission of epistemic goods, and (3) enable them to respond appropriately to the specific commitments of their audience (Hinchman Reference Hinchman2014). In cases where the epistemic inequality between the testifier and addressee is significant, as in the case of experts, these commitments become more pronounced, necessitating new competences.
What commitments and virtues might be involved in this scenario? Byerly’s account of intellectual dependability (Byerly Reference Byerly2021) could be helpful. In particular, Byerly distinguishes three groups of intellectual virtues that enable experts to promote epistemic goods in others via successful testimony. First, sincerity, honesty, and transparency facilitate sharing one’s perspective on topics relevant to others’ inquiries (Byerly Reference Byerly2021: 105). Second, communicative clarity eliminates potential sources of confusion (Byerly Reference Byerly2021: 127). Third, audience sensitivity involves paying attention to the unique characteristics of one’s audiences and tailoring communications to them (Byerly Reference Byerly2021: 144).
In summary, to navigate the epistemic social network, we rely on second-order evidence concerning potential testifiers, which we can capture through the concept of epistemic profiles. These profiles encompass one’s body of beliefs, categorized across different domains and evaluated based on the level of grounding for each belief. Additionally, they include epistemic virtues (both reliabilist and responsibilist). These virtues serve as markers for reliable testifiers, not only those more likely to acquire knowledge but also those with the ability to effectively transmit it.
Last but not least, we should expect epistemic experts to be epistemically honest. The honesty is situated on two levels: first, it is self-sincerity. We want experts to genuinely and reliably assess their epistemic position with respect to a proposition. Second, we expect them not only to assert their beliefs but also to be frank about their epistemic standing regarding those beliefs (see Ballantyne Reference Ballantyne2019). In particular, we expect experts to transmit their knowledge and not to transmit what they ascertain not to be knowledge, or a belief deemed good enough in the given context, or at least to acknowledge the reliability of the beliefs they testify to (Whitcomb et al. Reference Whitcomb, Battaly, Baehr and Howard-Snyder2017). That is why epistemic humility appears to be a very important virtue for experts, especially in socially and politically fragile environments where much is at stake (Zagzebski Reference Zagzebski, Pritchard, Farina and Lavazza2024).
Of course, it would be inefficient in epistemic economics if each of us had to individually create such profiles for fellow agents. This is where the concept of epistemic reputation enters the picture. Epistemic reputation so construed is a socially created, transmitted, and updated epistemic profile of individual agents, groups, and institutions. Thus, I can propose a definition of epistemic reputation that complements the earlier function (one and the other together build the theory of epistemic reputation):
Epistemic Reputation is social standing of agents, groups, and institutions based on their epistemic profiles in the given community.
5. Misleading reputation
What about cases where epistemic reputation is undeserved or grossly misaligned with an individual’s actual epistemic standing? A good illustration that likely also explains concerns about granting more significant epistemic entitlements to epistemic reputation might be a slightly developed case of President Dump originally proposed in Simion (Reference Simion2024b). In Simion’s case, Bill is a stubborn supporter of President Dump. He disregards all evidence in favor of Dump being a disastrous president. Consequently, he also disregards the track record, particularly instances where he trusted Dump’s opinions and decisions that turned out to be incorrect. A critic of epistemic reputation might add to this story that President Dump’s empirical reputation persists despite the fact that it fails to fulfill its intended function. In the remainder of this section, I will examine this and related counterexamples in order to clarify when and what kind of epistemic entitlements epistemic reputation gives rise to.Footnote 15
5.1. Function, malfunction, and normal functioning
The first step is to distinguish between functioning and malfunctioning systems, and to determine when a function is operating normally – hence normatively – and when the context renders appeals to reputation unjustified due to a distorted or inappropriate background. In this subsection, I draw on Graham’s distinctions (Graham Reference Graham and L.R.G.2023: 250–251). In the next one, I expand upon them, compare epistemic reputation with other cognitive mechanisms, and assess the extent to which misleading or distorted reputational signals should be a source of our concern.
First, Graham emphasizes that an item can malfunction temporarily or permanently, but it retains its function. For instance, a broken hammer still has the function of nailing things, and a malfunctional heart still has the function of pumping blood. Moreover, it is because of this function, which an item still has, that we say that the item is broken. Various other dispositions can still be fulfilled, but if the given function is not met, we say that the item does not work. Conversely, an item can be damaged in various other ways, but as long as it performs its function properly, we consider it to be working. As Graham reiterates, a function is not simply one of the dispositions of an item. It is the essential disposition of an item – be it a practice, institution, or procedure – that both confers meaning and establishes an evaluative standard. Accordingly, if we claim that the function of epistemic reputation is to signal the epistemic standing of agents, we are saying that the very reason we have epistemic reputation is precisely to fulfill that role.
Of course, epistemic reputation can play other roles as well. For instance, I may wish to honor a guest speaker at my conference according to their status, or a university might appoint academic ambassadors to engage potential sponsors, relying on their recognized achievements and standing. These uses, however, are contingent on the primary function of reputation, which is to signal epistemic standing. The university, in this example, seeks out scholars of genuine and recognized quality – those whose reputational standing effectively identifies them as such. In other words, these secondary uses presuppose the successful performance of reputation’s essential epistemic function. Conversely, there are cases in which epistemic reputation misleads, even if it may still fulfill other functions. But such instances are not arguments against the institution of reputation per se. Rather, they are examples of malfunctioning reputation. It is precisely by clearly articulating the function of epistemic reputation that we gain the normative basis for distinguishing bad cases of reputation from good ones. The normative concept of reputation underwrites our expectations about what should count as epistemically valuable reputation.
Graham further distinguishes between function and normal functioning. An item can still function properly even if, due to extrinsic circumstances, it cannot currently fulfill its function. A car on a lift, for instance, may be in perfect mechanical condition, yet unable to perform its function of transport. Likewise, the reason we have and transmit reputation in society, and the reason we take it into account in our epistemic practices, is its evidential function: reputation provides a socially distributed means of conveying epistemic profiles within a community. However, in epistemically polluted environments – those fed by distorted or false evidence – reputation may perpetuate inaccurate epistemic profiles. Thus, one may encounter a form of reputation that is functioning internally but, because of external distortions, no longer performs its evidential function reliably. Yet again, this is not an argument against the mechanism itself. Like other cognitive processes, reputation is epistemically valuable only within the appropriate context.Footnote 16 This point merits further elaboration, which I undertake in the following subsection.
5.2. Agent-dependent and input-dependent factors relevant to normal functioning
Earlier, in Section 3, I claimed that the function of epistemic reputation is to signal epistemic standing of agents within a community. Then, in Section 4, I clarified that reputation achieves this by tracking social standing, which itself is based on individual epistemic profiles. I can now specify that the successful functioning of epistemic reputation depends on two distinct factors.
First, reputation is not self-generated. It is based on underlying data: individual epistemic profiles. At a first approximation, reputation will only be as accurate as the input it receives (see Simion Reference Simion2024b). If it is fed false or distorted data, it will inherit their inaccuracy.
However, this is only part of the story. A second condition is equally crucial: the existence of a properly functioning mechanism for filtering, falsifying, and updating epistemic profiles. Ordinarily functioning epistemic agents – and by extension, epistemic communities – will revise reputational judgments in light of agents’ observable track records. Ideally functioning updating mechanisms would, in principle, be capable of accommodating even flawed initial inputs. False profiles would eventually be corrected through the cumulative evaluation of successes and failures. While such ideal rationality may be beyond our reach, relatively well-functioning updating mechanisms are sufficient for epistemic reputation to fulfill its function.
From this perspective, concerns about the reliability and validity of epistemic reputation can be reduced to broader normative questions in epistemology. On the one hand, all our cognitive mechanisms are fallible. Yet so long as the world is not epistemically demonic – meaning a world in which evidence would be systematically misleading – our cognitive mechanisms (perception, memory, testimony, inference, and epistemic reputation) remain good enough to yield knowledge. However, if we found ourselves in a world sufficiently close to a demonic one, no cognitive mechanism would be reliable – including those more basic than reputation.Footnote 17 The broader point is that our cognitive mechanisms (both those that directly produce knowledge and those support knowledge production, such as reputation), although imperfect, are still adapted enough to the degree of imperfection of our world.
On the other hand, just as even fallible environments can support knowledge, so too can imperfect epistemic agents. What we require – both for reputation and for basic cognitive mechanisms – is a minimal level of rationality, which consists, at its core, in appropriately responding to evidence (see Simion Reference Simion2024a: Chapter 6). A cognitive mechanism may malfunction either due to environmental distortions (e.g., lack of or polluted evidence) or due to the epistemic corruption of agents – that is, when agents fail to meet basic epistemic norms. This malfunction can be individual or collective, as illustrated in Dump-type cases.
Importantly, this is not a problem unique to epistemic reputation. Any cognitive mechanism can fail under such conditions. Reputation, although a form of second-order evidence, is generated and sustained by a collective cognitive process that relies on first-order evidential norms. When epistemic agents themselves are corrupted, the malfunction lies not in the mechanism of reputation per se, but in the agents who fail to maintain and revise it appropriately. Ultimately, therefore, this is again a broader epistemic problem.
That being said, as long as we do inhabit an epistemic environment that isn’t degenerate enough to impede other usual cognitive mechanisms, reputation remains reliable enough for its intended function.Footnote 18 Its performance, like that of other cognitive tools, is ultimately vindicated by its role in the broader social economy of survival – where effective cognition is not merely valuable, but indispensable.
My goal in the paper is to explain why, under normal circumstances, epistemic reputation translates into epistemic entitlements. Everything will hinge on whether a given case represents what is meant by normal circumstances. I argue that one of such constraints is that the relevant community (taking into account its fallibility) adheres to basic epistemic norms in principle. The case of Dump’s followers exemplifies radical disregard for basic epistemic norms. At its core, Bill (and other contributors to Dump’s reputation) fail to respond adequately to evidence (not just a specific second-order evidence, epistemic reputation). Under these circumstances, reputation malfunctions, akin to the defective heart from Graham. While developmental defects prevent the heart from performing its function, it doesn’t lose its proper function altogether. Regarding Dump’s followers, epistemic reputation doesn’t produce knowledge.Footnote 19 Primarily, it is because it erroneously elevates the epistemic status of the epistemic source (President Dump as a testifier), leading Bill to unjustifiably assume that testimonial belief is properly warranted. However, the sheer fact that reputation might malfunction does not itself mean that its function is different than assumed.
5.3. Reputational warrant and knowledge
Thus far, I have argued that when certain environmental conditions are met – most crucially, when the epistemic agent is sufficiently competent and the environment, while imperfect, is not profoundly polluted – epistemic reputation functions as a knowledge-supporting cognitive mechanism. It does so by supplying second-order evidence concerning the context-dependent reliability of testifiers. Otherwise, as a malfunctioning mechanism, it loses its epistemic entitlements. This raises an important question: what about the agent themself? Under what conditions is it epistemically permissible for an agent to rely on reputation? In other words, what warrant does reputation provide, and to what extent does this warrant license epistemic reliance?
To address this issue, compare once more reputation with other classical knowledge-generating processes. Perception, for instance, is a process that aims to produce accurate perceptual beliefs. Normally functioning perception consistently yields reliable perceptual beliefs. Therefore, prima facie, I have an epistemic entitlement to believe if a belief arises from perception. However, this doesn’t negate the fact that perception isn’t always reliable. In the brain-in-a-vat scenario, even though perception doesn’t consistently produce true perceptual beliefs, the agent is still justified in trusting it. This aligns with another element of Graham’s theory. I’ve proposed above that epistemic reputation serves as an epistemic warrant grade-one, that is, a warrant that a belief enjoys if and only if it is based on a (normally acquired) normally functioning belief-forming process, and if this process forms true beliefs as a function and reliably forms true beliefs in normal circumstances through normal functioning. It is still a warrant, providing justification (for instance, in demonic scenarios), but it doesn’t warrant knowledge on its own. Consequently, the epistemic entitlement I propose for epistemic reputation is ultimately quite weak but sufficient for justification.
Let’s consider a slightly modified case of Taliban (see Battaly Reference Battaly, Mi, Slote and Sosa2016).
SECT: Alec grew up in a secluded community, isolated from other sources of evidence. His entire family and natural surroundings revere the sect’s guru in a quasi-divine manner, treating him as the ultimate authority on all matters. Based on this reputation, Alec holds a strong conviction that the sect’s guru possesses the highest level of expertise, which undermines any opposing evidence. The guru systematically ventures beyond his area of expertise and provides incorrect answers to Alec’s questions. However, the questions posed are of such a nature that there’s no direct, empirical method of verifying them, except through the testimony of others. Nevertheless, this evidence is undermined by the epistemic reputation of the guru.
Indeed, guru’s epistemic reputation is malfunctioning. It mischaracterizes the guru’s epistemic status, thereby unduly inflating the weight of his testimony and disrupting the evidential balance. However, Alec continues to form justified beliefs. While he has no way to conclude that he’s been placed in an abnormal environment or to accept counterarguments, since epistemic reputation is prima facie justified second-order evidence, he still has justification (albeit for erroneous beliefs).
The key distinction between the Dump and Sect cases lies in the varying approaches taken to evidence that undermines reputation, which could potentially alter reputation itself. Simion (Reference Simion2024b) provides a helpful distinction between different scenarios of evidence rejection. She highlights that it’s possible to encounter profoundly polluted epistemic environments where it’s justified for an agent to mistakenly reject evidence. These situations should be distinguished from instances of evidence resistance. Considering the distinct origins of the reputation on which Bill and Alec rely, different level of social isolation, and the presence of strong defeaters, it’s plausible to diagnose that in the case of Sect, but not Dump, relying on reputation is justified. While both cases involve misleading reputation, Alec’s case involves an otherwise well-functioning epistemic agent, whereas Bill’s case involves an agent who systematically violates epistemic norms by disregarding readily available evidence. Furthermore, in Dump’s case, misleading reputation is not a contingent, agent-independent condition (like in Alec’s case), but rather a consequence of a malfunctioning epistemic agent (along with similar agents). Thus, Alec and Bill may be contrasted in reference to input-dependent and agent-dependent factors introduced above. Alec functions correctly but receives flawed input,Footnote 20 while Bill malfunctions and also receives flawed input (and partially contributes to it himself).
5.4. Abuses of reputation
In Subsection 5.1, I drew on Graham’s important distinction between the etiological or constitutive function of an entity and its possible contingent functions. For instance, the constitutive function of a stapler is to fasten sheets of paper together. Yet fulfilling this function often requires the stapler to be weighted and stable – features that enable it to perform other, secondary functions, such as holding down loose papers against a draught. Similarly, I have claimed that the constitutive function of epistemic reputation is to signal the epistemic standing of agents, even though it may serve other, non-epistemic purposes as well. But perhaps I have moved too quickly in affirming this function as normatively central. I address this concern in the present subsection.Footnote 21
Indeed, epistemic reputation often signals more than epistemic competence. High epistemic standing may also correlate with, or even act as a proxy for, non-epistemic attributes such as educational background, socioeconomic status, origin, or access to institutional privilege. In a society marked by epistemic injustice, the link between epistemic and social standing can become a mechanism for perpetuating racial, gender-based, and class-based prejudices. This raises a critical question: is the epistemic function of reputation sufficiently clear and robust to support its use as second-order evidence, as I have argued?
In response, we must distinguish between two versions of this worry – one benign, the other more corrosive. In the benign version, the objection is that epistemic reputation – while still tracking epistemic standing – also correlates with non-epistemic forms of social position (e.g., race, gender, class). Here, reputation functions as intended: it signals epistemic characteristics. However, these characteristics may incidentally support other social categorizations. While this may reflect broader structural injustices (such as unequal access to epistemic resources), the epistemic function of reputation remains intact. In this case, the injustice lies in the background conditions – specifically, the unequal distribution of epistemic goods and access – rather than in the reputational mechanism itself. I will return to this in the next section. For now, suffice it to say that this objection, though morally serious, does not fatally undermine the epistemic warrant that reputation can provide.
The more serious concern arises when epistemic reputation is distorted – not merely incidentally, but through systemic manipulation or the improper entanglement of epistemic and non-epistemic factors. Here, we must distinguish two layers of analysis:
First, at the situational level, such cases constitute malfunctions of epistemic reputation. As discussed in Subsection 5.2, this malfunction may stem from flawed inputs (erroneous epistemic profiles), or, more seriously, from corrupt evaluative mechanisms that fail to filter, revise, or update reputational data appropriately. In either case, the epistemic reliability of reputation is compromised.
However, and crucially, such cases do not invalidate the general epistemic entitlement of reputation. Despite our imperfect world, epistemic reputation remains, under broadly normal conditions, a sufficiently reliable mechanism for tracking the epistemic status of individuals, groups, and institutions. If it were not, we would not – and could not – depend on it as we do. The reason we trust it is that, under normal conditions, it performs its intended function; and since our conditions are statistically close enough to normal conditions, reputation can still be a valuable social cue of others’ epistemic positions.
Moreover, without good cases of reputation being fundamental ones, bad cases would also be inexplicable (see Graham Reference Graham and L.R.G.2023: 268). If reputation systematically misled us, we wouldn’t trust it.Footnote 22 The success of a manipulative reputation – such as the case of Dump, whose epistemic status is artificially inflated – depends on the general reliability of the reputational system. If epistemic reputation were systematically misleading, it would cease to function as a basis for epistemic trust, and we would cease to take it seriously. Deception, in this sense, parasitizes on the credibility of a generally trustworthy mechanism.
This structure is well illustrated by a non-epistemic analogy: the reputation of wine. Wine rating systems are designed to signal quality to consumers. That is their constitutive function and the reason for their existence. Likewise, the reason why customers rely on these ratings is because they need a proxy for quality – whether to choose a suitable gift, ensure a wine is appropriate for an occasion, or determine whether a price is justified. Now, suppose a corrupt reviewer inflates the scores of certain producers. While this clearly contravenes the intended function of the rating system, it does not immediately undermine the consumer’s rationale. The consumer is still justified in consulting the ratings because of the system’s assumed function. Indeed, the deception only makes sense to the manipulator insofar as consumers continue to trust the ratings.
The fraudulent use of reputation parasitizes on its proper, cognitive function. So long as the manipulation is subtle – meaning that, in a sufficiently large number of cases, the reputational rankings still correlate with actual wine quality – the system remains functional. Consumers tolerate a margin of error – which may be sufficient to conceal occasional manipulation – because, on the whole, the ratings remain informative. But if the manipulation became pervasive and reputation systematically failed to track quality, the system would collapse. The same is true for epistemic reputation: its credibility depends on its general reliability, not perfection.Footnote 23
However much Dump may scare us, epistemic reputation still generally works well enough. Like all fraudsters, cases like this one parasitize legitimate cases of trustworthiness. Eventually, the level of abuse of epistemic reputation, like other trust-related evidence, might be so great that it would undermine the very rationality of relying on it as second-order evidence. One option would be to say that this kind of evidence is too weak to generate relevant epistemic entitlements, and require some further justifications, but this would be an epistemically costly solution. As a society, we’re willing to pay the price of higher non-reliability because this evidence is too valuable to us (Levy Reference Levy2022; Mercier Reference Mercier2020).
6. Responsibility for reputation
Before concluding, I would like to briefly address two broader issues that arise from the preceding analyses. On the one hand, these reflections serve to situate the discussion of epistemic reputation within wider debates in social epistemology. On the other hand, they are not intended as definitive resolutions, but as prompts for further inquiry – openings rather than closures. The account of epistemic reputation developed thus far brings into view several pressing questions that concern both individual and collective epistemic responsibility. I will discuss them in turn.
6.1. Responsibilities on behalf of society
Earlier, I defined epistemic reputation as the social standing of agents, groups, and institutions. As a social entity, it differs from the actual epistemic position it aims to convey. Therefore, one of the measures of reputation’s fittingness lies in its correlation with its subject’s track record. Since reputation is expected to help in navigation within social epistemic networks, and practical interests depend on this navigation, even if the subject of reputation for various reasons (fame, profits, and personal gains) holds a higher position than deserved, it is in the community’s interest to calibrate reputation to reflect their actual epistemic standing. A consequence of the reputation function is its self-regulation.
If reputation is so good, why are we reluctant to grant it a stronger epistemic status? I argue that on a purely epistemic level, reputation is a self-regulating social entity. To explain its unreliability, we must consider external factors that shape epistemic evaluation by interfering with the construction of epistemic profiles. One such factor is prejudice arising from stereotyping. The social position of epistemic agents, groups, and institutions is significantly influenced by gender, racial, ethnic, economic, political, and other factors. In a given community, an actual expert may go unrecognized or have their position downplayed because of their gender, age, race, ethnicity, or political affiliation. For instance, a woman’s testimony may be downgraded due to alleged bias, even if she is an expert. Additionally, the intellectual competence of individuals who are not fluent in a foreign language, speak with an accent, are dyslexic, or have speech and language disorders can be underestimated (judging intellectual performance from speech performance). As a result, just like in Dump case, reputation can wrongly overestimate the epistemic standing of individuals and groups, it can also underestimate it, particularly, if tainted by prejudice.
However, these examples are not an argument against epistemic reputation. Rather, the epistemic significance of reputation explains why epistemic injustice is such a serious epistemic problem (M. Fricker Reference Fricker2007; Medina Reference Medina2013). Precisely since we rely on reputation in testimonial interactions (Begby Reference Begby2021; Levy Reference Levy2021; Mercier Reference Mercier2020), systemically entrenched epistemic injustices have such a profound impact. As long as reputation remains crucial second-order evidence for testimonial knowledge, we, as a community, must take on a serious responsibility for the moral and epistemic task of combatting various forms of epistemic injustice. Because of the epistemic impact of reputation in testimonial exchanges, the harms that epistemic injustice inflicts affect not only the personal goods of the victims but also significantly weaken our epistemic environment (Gunn Reference Gunn2020; Jarczewski Reference Jarczewski2025a), leading to further (systematic) epistemic damage.
6.2. Responsibilities on behalf of experts
So far, I’ve focused on society’s obligations regarding the recognition of an agent’s, group’s, or institution’s epistemic position based on their track record. However, we can also consider the perspective of the subject of reputation. Earlier, I mentioned that an expert can only fulfill their social role to the extent that they are recognized as such. Doesn’t this then impose additional duties on the expert to make themselves known? Our intuitions on this matter are intricate. Here, I would like to suggest what a study in that matter would involve. Consider the following case:
CARDIOLOGIST: Ann is a cardiologist:
-
1. On an airplane, a passenger faints. The flight attendant asks if a doctor is present.
-
2. Susan, an avid collector of random information, has some medical questions. Bored at a party, she asks if someone could answer them for her.
-
3. Ann is suddenly called for an operation. In a Facebook group of neighbors, Henry is looking for someone to help his son with a biology question he needs to answer for an upcoming exam in an hour.
-
4. The mob boss seeks advice on how to finish off an opponent under the guise of a heart attack.
Clearly, in case 1, Ann has a moral obligation to identify herself as an expert. In case 4, she has a moral obligation not to reveal herself. For various reasons, it seems plausible that she has no obligation to reveal herself in cases 2 and 3.
Extending this concept to the systematic creation of one’s reputation, we encounter two opposing intuitions. On the one hand, it appears that epistemic humility is a genuine and undervalued epistemic virtue (Roberts and Wood Reference Roberts and Wood2007). Excessive self-promotion could be perceived as contradictory to this virtue. On the other hand, recognizing experts positively impacts the quality of the social epistemic environment. Appropriate self-signaling is one way to enhance this environment. So one should do all their best to make them known as expert.
I think it’s possible to connect these intuitions, especially if we view reputation as a multi-level epistemic profile of individuals, groups, and institutions. A crucial aspect that contributes to a proper reputation is signaling an agent’s limitations and shortcomings. This aligns with epistemic humility (Ballantyne Reference Ballantyne2019; Whitcomb et al. Reference Whitcomb, Battaly, Baehr and Howard-Snyder2017; Zagzebski Reference Zagzebski, Pritchard, Farina and Lavazza2024). Therefore, an expert’s concern for a proper reputation extends beyond being motivated to avoid underestimation; it also involves ensuring that their position is not overestimated either. This is evident in the practice of putting one’s reputation at stake.Footnote 24 A prudent expert avoids epistemic transgressions, knowing that society severely punishes such actions (Dasgupta Reference Dasgupta and Gambetta1988; Goldberg Reference Goldberg2015: 210; Mercier Reference Mercier2020: 86–90). Since the function of reputation is to accurately mark one’s epistemic standing, and only indirectly contributing to knowledge, experts who signal that their position is lower than we would expect are gently downgraded, according to their auto-evaluation. Conversely, if we discover that they have inflated their reputation, by claiming to be more competent that they are or by breaking epistemic norms of assertion, we punish them more severely. This downgrading is not solely based on their actual competence but also reflects a lack of honesty and self-awareness in assessing one’s limitations. We no longer trust their own testimony on their epistemic position and the fact of epistemic trespassing is a second-order defeater for their future testimony.
There is a need and a social expectation for experts (and others) to contribute to being accurately evaluated epistemically by the community (i.e., neither too low nor too high). Certainly, contributing to one’s proper reputation is beneficial to the society. Does it follow from the fact that such a self-signaling is an (epistemic) good that there is an associated (imperfect) duty? This topic deserves further exploration in a separate paper. Here, I introduce these themes and demonstrate how they connect to the broader theory of reputation.
7. Conclusion
In my paper, I argued that the function of epistemic reputation is to signal the epistemic position of different agents within the epistemic community. As such, reputation serves as crucial second-order evidence on the epistemic standing of individuals, groups, and institutions for the social division of epistemic labor. This function explains the existence and position of reputation even in the face of its misuse. Drawing from Graham’s distinction between grades of warrants, I explained why reputation can, in certain polluted environments, provide justification (for testimonial beliefs based on information about reputation), while simultaneously failing to generate knowledge. Reputation is not immune to biases and stereotypes, which can account for various forms of epistemic injustice. The recognition of its significant epistemic role is essential for comprehending why these injustices constitute such a grave epistemic (and not only moral) evil. In the concluding part, I emphasized the additional responsibilities that society and experts bear in constructing a well-calibrated reputation.
Acknowledgments
I want to express my deepest gratitude to Sandy Goldberg for his invaluable feedback on the previous versions of the paper. I also want to thank Chris Kelp for his encouragement to pursue this subject. Additionally, I am grateful to Wayne Riggs, Stephen Grimm, Peter Graham, and the participants of the graduate seminar at Fordham University, the 25th World Congress of Philosophy in Rome, and the 4th Social Epistemology Network Event for their insightful comments on the earlier versions of the paper and related research.
Funding
This research was funded by the National Science Centre, Poland, project no. 2021/43/D/HS1/00977.
Competing interests
The author declares no conflict of interest.
Open access
