12-DAIMON-93-(web)

On testimonial justice online. Nuancing Karen Frost-Arnold’s optimistic virtue epistemology

Sobre la justicia testimonial online. Matizando el optimismo de Karen Frost-Arnold acerca de la epistemología de la virtud

GONZALO VELASCO ARIAS*

Recibido: 12/04/2024. Aceptado: 27/06/2024.

* Gonzalo Velasco Arias es Profesor de Filosofía en la Universidad Carlos III de Madrid. Sus principales líneas de investigación son (1) la aproximación filosófica a las emociones políticas, (2) las injusticias epistémicas en el espacio público digital y (3) el análisis de las manifestaciones subjetivas de las injusticias estructurales. En relación a estos temas, recientemente ha publicado el artículo “Arrogancia, desconfianza y desafección. Una aproximación desde la epistemología de la virtud”, Isegoría. Revista de Filosofía moral y política, N.º 70, enero-junio, 2024, https://doi.org/10.3989/isegoria.2024.70.1348; el ensayo Pensar la polarización, Gedisa, Barcelona, 2023; así como la coedición junto a A. Gómez Ramos del volumen colectivo Atlas político de emociones, Trotta, Madrid, 2024. Es miembro de los proyectos de investigación “El vínculo y su contrario. Desafección, mediaciones y representación política (VI_CO)” (PID2021-124954NB-I00), “Desacuerdos morales en la esfera digital. Dinámicas interactivas, micromecanismos y marcadores culturales (digi_morals)”, Fundación BBVA. gvelasco@hum.uc3m.es.

Abstract: In Who Should We Be Online, Karen Frost-Arnold advocates an approach to epistemic virtues that resists pessimism about the possibility of our online epistemic agency being responsible and socially just. On the basis of a veritist epistemology, her proposal overcomes both responsibilist individualism and the socio-structural critique that delegates all responsibility to institutional transformations. The author identifies in online lurking an activity unique to online epistemic agency that can provide exposure to messages from people discriminated against by epistemic injustices. For Frost-Arnold, moreover, this implies the possibility of the lurker experiencing epistemic frictions that will favour a more reliable willingness to be fair in giving credit to the testimonies of those discriminated against. In this note I will qualify this optimistic stance, arguing the epistemic individualism that underlies it. I will point to a group virtue model as a possible solution.

Keywords: epistemology of virtue, online testimony, deference, lurkers, trust, humility.

Resumen: En Who Should We Be Online, Karen Frost-Arnold defiende una cierta aproximación a las virtudes epistémicas que resista al pesimismo acerca de la posibilidad de que nuestra agencia epistémica online sea responsable y justa. Sobre la base de una epistemología veritista, su propuesta supera tanto el individualismo responsabilista como la crítíca socio-estructural que delega toda responsabilidad en transformaciones institucionales. La autora identifica en el online lurking (fisgoneo online) una actividad exclusiva de la agencia epistémica online capaz de proporcionar una exposición a mensajes de personas discriminadas por injusticias epistémicas. Para Frost-Arnold, a su vez, esto implica la posibilidad de que el lurker experimente fricciones epistémicas que favorecerán una disposición más fiable a ser justos a la hora de dar crédito a los testimonios de personas discriminadas. En esta nota matizaré esta postura optimista, alegando el individualismo epistémico que subyace. Apuntaré a un modelo de virtudes grupales como posible solución.

Palabras clave: epistemología de la virtud, testimonio online, deferencia, fisgones, confianza, humildad.

1. Introduction

Online epistemic virtue is the focus of Karen Frost-Arnold’s enquiry in Who Should We Be Online (Frost-Arnold, 2023). The research objectives of the book can be synthesised in the attempt to outline the conditions under which responsible epistemic agency can take place in the hybrid media space (Chadwick, 2017), including in this consideration of responsibility the dimension of epistemic justice. Far from indulging in a naïve conception of virtue, her enquiry embeds an evaluative reflection on the epistemology of virtue that allows her to refine a model applicable to the analysis of online agency. In my view, this takes the form of two issues, which I outline as an introduction to the rest of my argument.

First, Frost Arnold eschews the nostalgic metanarrative that argues that the epistemic and informational problems of our hybrid media system can be solved by recovering or enhancing classical virtues. Instead, she advocates an epistemology of situated virtue that allows for a non-ideal approach to the normative issues that arise from our online epistemic agency. In addition, these situated virtues are framed in a systems-oriented social epistemology that considers that what matters is not identifying which virtues are useful for the risks inherent in online epistemic agency (Heersmink, 2018; Vallor, 2016) (Driver, 2001, p. 68), but identifying which vices and which virtues are fostered by the epistemic structure of the network. This, again, entails a response to the exceptionalist narrative. This view redresses the motivationism and doxastic voluntarism behind the responsibilism that is present in the social discourse and in many approaches that still understand education and training in offline epistemic virtues as the only way to improve epistemic behaviours and to mitigate the intrinsic risks of online communication (Heersmink, 2018, 10).

Secondly, the combination of virtue epistemology with a certain systems-oriented social epistemology advocated by Frost-Arnold enables to overcome the dichotomy between individual responsibilism and socio-structural critique (Anderson, 2012). Frost Arnold’s work is an attempt to safeguard agency without yielding to naïve faith in individual virtue or to the maximalism of socio-structural critique.

Despite this sophisticated version of epistemic virtue, I believe that Frost-Arnold’s normative proposals for epistemic agency online in relation to epistemic injustices remain grounded in a certain epistemic individualism. I will articulate my critique with reference to chapter 5, devoted to virtuous lurking. According to the author, virtuous lurking is a specific practice of online epistemic agency, which renders the exposure of privileged subjects to epistemic frictions and unlearning easier through the reception of testimonies issued by users belonging to objectively discriminated collectives, while avoiding the undesired effects of the interference of privileged agents in the conversation between discriminated agents. Frost-Arnold argues that this allows for hopeful trust on the part of marginalised subjects. The possibility of these practices is thus a reason for optimism about the possibility of more responsible agency in relation to epistemic injustices online. In order to elaborate my account of Frost-Arnold’s optimism, I first need to make some analytical clarifications about (1) approaches to online testimony, (2) the account of testimonial justice as a virtue, and (3) some specific phenomena of the digital epistemic environment that hinder responsible and justice-focused agency.

2. Testimony and online deference

In online or hybrid communicative contexts, restrictive definitions of testimony and deference are not useful. Narrow definitions of testimony are those that only accept as such speech acts in which the speaker intends to present evidence to someone about a matter known to be in dispute. Equally narrow is the view that limits testimony to speech acts in which the speaker claims that his statement is true, thereby committing the hearer to believe and trust him. Given the complex interactivity and ambiguity of speech acts in hybrid or online communicative environments, I believe it is appropriate to accept a permissive approach to testimony (E. Fricker, 1995), which with Sosa could be defined as “a statement of someone’s thoughts and beliefs, which they might direct to the world at large and not one in particular” (Sosa, 1995, p. 219). Sullivan has argued that, on the Internet, specifically on social networks, although a statement can be reposted indefinitely without the original message being modified, the communicative meaning can change from the original (Sullivan, 2019). The consequence of this is that not even the informational basis of the testimony is preserved, which would be the condition required by approaches such as that of Lackey (Lackey, 2007). Therefore, strictly speaking, we could not speak of “testimonies” in digital environments.

Despite this, digital environments are in fact used as a source and means of expression of information and testimonies. Frost-Arnold provides a wide range of evidence of communicative situations in which discriminated groups use their online avatars to express and testify about the injustices they suffer. Regardless of the communicative acts they use to do so, Frost-Arnold sees this phenomenon as an extension of the production and circulation of true beliefs about situations of injustice, and as an opportunity for those who, because of their position of privilege, do not have access to these life experiences.

Although it is not a concept employed by Frost-Arnold, I believe that the problem of deference is implicit in reflection on the epistemic friction to which the Internet exposes privileged subjects. Heuristically, I think it is convenient to adopt a broad definition of deference, in the sense that “A defers to B on the question whether p if A believes that p (or not-p) merely because B believes that p (or not-p)” (Brinkmann, 2022, p. 267). This definition does not necessarily require either that there be intentional testimony or that the speaker be an expert in the domain of beliefs about which testimony is given, so it fits well with the notion of online testimony that I have just defined, and with the kind of listening that occurs to those who have testified to the suffering of epistemic injustice. Deference is a radical act of trust in which the listener assumes an epistemic risk that implies a strong normative expectation in the trustworthiness and benevolence of the speaker. In other words, from the point of view of persuasive argumentation, it is irrational for a privileged subject to trust the testimony of a marginalised subject and eventually accept that this testimony forces him to change his own set of beliefs. I suggest thant when Frost-Arnold identifies online lurking as an opportunity for privileged agents to “unlearn” their prejudices and biases, he is not taking into account this excessive cost..

Deference in contexts of epistemic injustice, moreover, presents a difference with respect to modalities of moral or epistemic deference in more defined fields of knowledge or practice. The motivations for deference can be twofold: (a) the speaker’s expression of his or her experience is convincing and persuades the listener to delegate the opinion to him or her, or (b) the listener recognises the testimonial and/or hermeneutic deficit to which the speaker is subject because of his or her group or social identity and, in order to compensate for this deficit, decides to trust him or her regardless of his or her agreement with his or her set of beliefs. In turn, testimony (in the broad sense referred to above) can be about the situation of epistemic injustice itself, or about a particular field of experience.

3. Is epistemic justice a virtue?

Virtue epistemology defines knowledge as a justified and non-accidental true belief that is a product of the agent’s reliable epistemic competence. Applied to situations of testimonial injustice, it is virtuous “the hearer who reliably neutralize prejudice in her judgments of credibility”. In other words, testimonial justice aspires to a state in which we do not too often find ourselves in the situation where we notice that we are committing unrectified epistemic injustices. In that case, we would be unreliable. For Fricker, self-corrective motivation is a necessary component for the reliability of the epistemically just agent. Although, in her responses to the criticisms raised by Shermann and Alcoff, she makes some concessions to a fallibilist version according to which it is possible to train and turn into habit the moments of apperception that motivate the correction of our epistemic vices (M. Fricker, 2010, p. 92). At the same time Fricker insists that such reliability can only be sustained by a hearer’s motivation to do justice. Thus, although Fricker’s own theory is an objection to doxastic voluntarism, there is a certain practical voluntarism.

The main criticisms of virtue epistemology come from “externalist” approaches such as situationism and socio-structural criticism. For socio-structuralism, the situations that impede the exercise of virtues are not contingent or trivial but regular and socially relevant, and have to do with structural power relations that shape our unconscious biases and prejudices. Hence, for Anderson, promoting individual virtues to redress structural epistemic injustice, while not wrong in itself, “plays a comparable role to the practice of individual charity in the context of massive structural poverty. Just as it would be better and more effective to redesign economic institutions so as to prevent mass poverty in the first place, it would be better to reconfigure epistemic institutions so as to prevent epistemic injustice from arising” (Anderson, 2012, p. 171).

The alternative to individualistic responsibilism, therefore, is often the institutional dimension. For Anderson, this does not exclude the promotion of individual virtue (Anderson, 2012, p. 166). However, from the point of view of the epistemology of virtue, the problem with the institutional solution is that if the application of virtue is conditional on institutional determinants linked to a concrete situation, then it no longer fulfills the condition of being cross-situationally consistent.

4. Arguments for pessimism

In my opinion, the main reasons for pessimism are the phenomena of information segregation, the spread of mistrust caused by the phenomenon of impostors and fakers, and also epistemic individualism that fosters “illusions of online understanding” (De Ridder, 2022). The question then is to what extent these virtuous practices have a normative force to overcome these difficulties. Since the first factor has been dealt with very abundantly by the specialised literature, I will focus on the last two.

4.1. The possibility of online imposture as a source of distrust.

Trustworthiness is a demand that the epistemic agent makes of the witness, to compensate for the risks associated with trusting. In trusting, the agent is vulnerable to intentional misinformation from the witness, and relies on the witness’s knowledge of the issue to accredit the trust placed in him or her. Trust, therefore, requires normative expectations: in order to be trusted, the witness is expected to be both competent and benevolent (Dutilh Novaes, 2023; Levy, 2022). It is this second expectation, benevolence, that the online communicative environment systematically betrays because anonymity enables the phenomenon of the impostor. Anonymity makes it difficult to accredit the authenticity of the witness. Excluding other philosophical approaches to identity, I will restrict here the meaning of “authenticity” to consistency. The authenticity of an online user can then be encoded in his or her consistency, and this can be assessed as (i) the coincidence between the online and offline self, (ii) the consistency in action and self-presentation across platforms, and (iii) the consistency in online presentation over time (Frost-Arnold, 2023, p. 83). Those who present themselves consistently in the online debate are presumed to be reliable in ther intentions and behaviour1.

Several authors have analyzed the distortions in trust generated by significant cases of impostors on various online platforms. Without going into the specificity of these cases, we can induce that the intrinsic possibility of online faking and imposture, the difficulty of verifying the authenticity of testimonies, and the very structure that fosters gaming and simulation, lead to the fact that, in conditions of vulnerability, mistrust in testimonies not accredited by authority or expertise is the default attitude (Frost-Arnold, 2023, pp. 77, 81). Online spaces are therefore hostile to the expression of testimonies of marginalisation and injustice. For this reason, it has been suggested that the major online social media platforms might modify their authenticity clauses, remove the right to anonymity or moderate it to pseudonyms whose authentic identity would be preserved by third party regulators who could apply progressive sanctions in case of abuse of the privileges of non-disclosure of authenticity (Véliz, 2019). However, it is also undoubtedly true that anonymity enables disruptive agencies that allow for the expression of testimony and political demands that under conditions of authenticity would be stifled by various forms of epistemic injustice. Other authors have argued for institutional markers of trustworthiness (Rini, 2017). However, these markers would increase epistemic gaps with those who do not have a prior trustworthy track record .

4.2. Online illusions of understanding.

From the point of view of research epistemology, although search engines are an undoubtedly agile tool for finding information, at the same time they feed “illusions of understanding” (De Ridder, 2022). Search engines, even more so in conversational versions such as ChatGPT, encourage confusing the mere connection of fragments of information with the process of comprehensive understanding (thus generating a false sense of self-sufficiency), pre-determine enquiry strategies with mechanisms for auto-completion of searches, impose arbitrary criteria for the evaluation of findings (order of appearance of information, popularity among users) and subject the agent to an infinite recursivity of information that makes it difficult to discern when sufficient evidence is available. This is the same effect caused by the availability and accessibility of expert opinion, as it can lead to a loss of track of our trustworthiness and distort our understanding of our own abilities (Fisher et al., 2015, p. 675).

The concept of “illusions of understanding” as applied to online enquiry raises two issues that need to be differentiated. The first (a.i) concerns perceptions of our own cognitive abilities and the fulfillment of our epistemic duties. By increasing our innate disposition to overestimate our abilities, it produces less reliable agents and, moreover, makes it difficult to correct this overestimation based on a certain amount of critical introspection. From a responsibilist point of view (i.b) “illusions of understanding” generate a false sense of fulfilling individual epistemic duties and thus foster less responsible agents through induced arrogance (Levy, 2019).

5. Lurking, optimism and the epistemic humility debate

To recapitulate, epistemic agency in hybrid online/offline environments tends to be more arrogant and individualistic due to ignorant overestimation of one’s own capacities, and also less trusting due to a propensity for imposture that makes it difficult to assess the authenticity of testimony. Faced with the difficulty of identifying evaluation criteria that justify deference in testimonies of marginalised subjects, a first temptation would be to ask once again about the possibility of appealing to the intervention of the subjective intellectual virtues of the listener that would allow him to defer reliably. From the systems-oriented epistemology approach employed by Frost-Arnold, what is interesting to know is whether the Internet is conducive to those dispositions that favour virtuous deference capable of compensating for the evaluative difficulty of online testimonies.

Karen Frost-Arnold identifies this possibility in the figure of the lurker. Lurkers are people who listen to, read or visualise expressions or communicative exchanges without participating in the communication themselves. The Internet, due to the guarantees of anonymity it provides, favours opportunities for this type of activity. The protection of anonymity would prevent defensive intuitive reactions that would arise in direct interpellation, and create the conditions for the benefits of “epistemic friction” or “world travelling” (Lugones), to mention some concepts that have been used to explain the possibility of the privileged learning about their own ignorance (Frost-Arnold, 2023, p. 174) by listening to the experiences and testimonies of discriminated groups. According to this view, lurking brings truthful benefits to the practitioner and avoids the unwanted side effects of other forms of well-meaning interaction of privileged allies with marginalised subjects. Above all, it would avoid what Sullivan has called “ontological expansion” (Sullivan, 2006) (the self-attribution by privileged agents of the right to participate in communicative scenarios that are modified by that participation), or what Bernstein has called “epistemic exploitation” (Berenstain, 2016), a phenomenon that occurs when the privileged alleged allies place the burden of proof and the responsibility to educate them on the shoulders of the discriminated and, in doing so, increase rather than decrease the epistemic gap.

Thus, if for veritism the reliability of a socio-epistemic practice can be measured by the ratio of true beliefs acquired to the total number of beliefs produced by that practice (Badhwar, 2009; Goldman, 2010), then lurking can be considered a reliable practice. The mere possibility of lurking would enable epistemic virtues such as open-mindedness, curiosity or humility. To my mind, there are two major objections to this view. (a) In my view, Frost-Arnold’s analysis does not sufficiently explain whether these are virtues enabled by lurking, or whether they are character traits necessary for lurking to be virtuous. If the latter, as I am inclined to think, the argument would nevertheless rest on a responsibilist voluntarism. (b) Although the struggle against epistemic injustice associated with an action such as lurking involves a set of distinct virtues, in my view epistemic humility is the meta-virtue shared by all of them. Humility, for many, is more of a meta-virtue, because it is a willingness to revise our epistemic beliefs and attitudes when new evidence or testimony presents itself (J. S. Baehr & Hazlett, 2016). It thus entails a motivation to recalibrate our capacities, skills and experiences presupposed in inquisitiveness, curiosity or open-mindedness (the willingness to transcend the default standpoint and to take into consideration the merits of other standpoints) (J. Baehr, 2011, p. 152). According to Frost-Arnold’s optimistic approach, epistemic life on the internet does not necessarily foster hubris but, by enabling specific dispositions such as those of the lurker, favours an intellectual humility that is not possible in offline environments.

In a very interesting twist of argument, Levy has argued that behind humility lies a presupposed arrogance: that of an epistemological individualism that assumes that we are always capable of autonomously revising our beliefs and that, ultimately, understands that only beliefs that have been subjected to critical examination qualify as knowledge (Levy, 2023). Whether for evolutionary reasons or because of enlightened cultural ascendancy, this epistemological individualism has become the normative common sense of our everyday epistemic practices. For Levy, this contradicts the interdependent nature of our epistemic agency, which allows him to paradoxically interpret experiments that have provided empirical evidence about the alleged epistemic hubris behind the “illusions of understanding” (de Ridder). These experiments induce intellectual humility by showing participants that they were actually ignorant of how mechanisms they thought they knew work, or of policy measures they had chosen to support. The overarching effect is a distrust of prior beliefs and the emergence in participants of a more humble disposition towards their own cognitive capacities and epistemic resources. Levy does not evaluate these findings positively. On the contrary, in his opinion this demonstrates an epistemological individualism that pushes to revise unjustifiably (because of lacking the competences to do so) beliefs formed through justified deference. The paradoxical conclusion of this reasoning is that humility, instead of favouring deference, may on the contrary neutralise previous acts of virtuous deference.

This reasoning does not apply well to situations of structural epistemic injustice in which, precisely, the appeal to humility implies revisiting a deference that has had discriminatory effects. Nevertheless, I do find the critique of the underlying epistemic individualism useful: regardless of whether the prior deference was virtuous (Levy discusses situations in which humility can lead to refuting confidence in scientific consensus) or flawed (because beliefs loaded with discriminatory biases are accepted), the conclusion to be retained is that confidence in the possibility of an act of individual contrition is illusory. In other words, the online lurking of testimonies of discriminated people is not a sufficient guarantee for an individual to revise his or her previous assumptions, reformulate his or her beliefs and identify his or her biases. In fact, taking Levy’s reasoning to the extreme, it could be irrational because it would push one to take excessive epistemic risks. This is why, in my view, trust and deference in testimony requires self-confidence that can only be provided relationally.

6. Coda: towards a model of group virtue?

My conclusion is that online lurking, although it may be a paradigm of how the Internet can trigger virtuous epistemic dispositions, does not have sufficient normative force to overcome the risks for the agent of trusting the testimony of subjects who, precisely because they suffer epistemic injustices, do not satisfy the requirements of trustworthiness. I believe that this conclusion authorises the leap to group virtues in search of a way to minimise the risks of this trust. This has been attempted by Lavinia Marin and Samantha Copeland, who have argued that self-confidence can only be born in communities that foster critical relational behaviour (Marin & Copeland, 2022). I believe that Mark Alfano’s conception of communities of trust (Alfano, 2016), whose members share the meta-knowledge of reciprocal trust, is also useful for thinking about this leap to the group level. Although Copeland and Marin understand relational trust as a practice and Alfano rather as a shared certainty about the expected behaviour of others, both approaches agree in arguing that for it to be rational to trust an out-group witness, in-group trust has to be warranted. More specifically, Copeland and Marin describe trust in those with critical attitudes. The idea would be as follows: “All members of group G trust that if an agent X has a reasonable critical attitude about the group’s position on issue A, and if he/she decides to trust the testimony B of an agent Y external to the group, the members of group G must a priori trust X’s good intention and X must know with certainty that the other members trust his/her action”. This does not automatically imply transitivity, in the sense that because of that trust, the members of G automatically come to believe Y’s testimony. This thesis would jeopardise the existence of the group. What this model seeks to guarantee is that the action of trusting and trying to change one’s own belief system does not have an excessive cost, neither from the point of view of persuasive argumentation theory nor from the point of view of the motivations linked to group membership.

Although the purpose of this paper is not to develop that model, I believe that for the online lurking advocated by Frost-Arnold to be effective from a veritist (to produce and spread more true beliefs) and ethical (to take responsibility for one’s own responsibility for certain injustices and commit to change oneself to mitigate them) points of view, group conditions must be in play to justify the rationality of these epistemic and ethical decisions. Otherwise, online lurking may be a way to alleviate the bad conscience of privileged actors at the cost of not triggering any commitment to transform themselves

References

Alfano, M. (2016). The Topology of Communities of Trust.

Anderson, E. (2012). Epistemic Justice as a Virtue of Social Institutions.

Badhwar, N. K. (2009). The Milgram Experiments, Learned Helplessness, and Character Traits.

Baehr, J. (2011).

Baehr, J. S., & Hazlett, A. (Eds.). (2016). The Civic Virtues of Skepticism, Intellectual Humility, and Intellectual Criticism. En

Berenstain, N. (2016). Epistemic Exploitation.

Brinkmann, M. (2022). In Defence of Non-Ideal Political Deference.

De Ridder, J. (2022). Online Illusions of Understanding.

Driver, J. (2001).

Dutilh Novaes, C. (2023). The (higher-order) evidential significance of attention and trust—Comments on Levy’s

Fisher, M., Goddu, M. K., & Keil, F. C. (2015). Searching for explanations: How the Internet inflates estimates of internal knowledge.

Fricker, E. (1995). Telling and Trusting: Reductuonism and Anti-Reductionism in the Epistemology of Testimony.

Fricker, M. (2010). Replies to Alcoff, Goldberg, and Hookway on

Frost-Arnold, K. (2023).

Goldman, A. I. (2010). Systems-oriented social epistemology. En

Heersmink, R. (2018). A virtue epistemology of the Internet: Search engines, intellectual virtues and education.

Lackey, J. (2007). Why we don’t deserve credit for everything we know.

Levy, N. (2019). Due deference to denialism: Explaining ordinary people’s rejection of established scientific findings.

Levy, N. (2022).

Levy, N. (2023). Too humble for words.

Marin, L., & Copeland, S. M. (2022). Self-Trust and Critical Thinking Online: A Relational Account.

Sosa, E. (1995).

Sullivan, Emily. (2019). Beyond Testimony: When Online Information Sharing is not Testifying.

Sullivan, S. (2006).

Vallor, S. (2016).

Véliz, C. (2019). Online Masquerade: Redesigning the Internet for Free Speech Through the Use of Pseudonyms.

1 As one reviewer has pointed out to me, users can be consistent in their inauthenticity. Indeed, consistency ii and iii do not imply benevolence or transparency of intentions at all. For this to happen, type 1 - consistency between the online and offline user - must also occur. And since, by definition, the current user model does not guarantee this requirement, I argue that anonymity is a structural source of distrust... unless progress is made in implementing some of the measures suggested by Veliz (2019), as I indicate below.