1. Taking virtuous acts as evaluatively and conceptually primary

Consider the following case. At a public event, you and your friend hear a speaker making a strong argument in favor of cutting social welfare programs to reduce the national deficit. You immediately reject the argument, feeling that it would harm those in need. However, your friend, who shares your concerns, decides to listen carefully and asks the speaker to elaborate on how they believe the cuts would benefit society. Your friend seeks to understand the reasoning behind the proposal, even though they don’t agree with it.

Do you say, “That was open-minded of you,” or “That was open-minded of you on the condition that it issued from a stable disposition to act from similar motives in similar circumstances” (Hurka 2016:71)? Clearly, you say the former. Our judgment of the virtuousness of acts turns on the occurrent motives and acts, rather than on whether the act issued from a stable trait of character.

In fact, your friend does not possess the virtue of epistemic open-mindedness. According to AFR, your friend’s act is virtuous because they have performed an open-minded act based on good epistemic motives. Proponents of the traditional view disagree. For them, your friend did not perform a virtuous act because they did not perform an open-minded act based on the praiseworthy character trait of open-mindedness. According to the traditional view, a virtuous act must proceed from a virtuous character trait; if it does not, the act might be one which an open-minded person would perform but is itself not virtuous (Aristotle 1980:1105a31–b12 and Hurka Reference Hurka2006:70). Whereas, on the traditional view, possessing a virtue, i.e., a virtuous disposition, is a necessary condition for an act to be virtuous, it is not necessary for AFR.

To draw a further distinction, imagine that your friend did not perform an open-minded act because they wanted to gain understanding, but rather because they were afraid to interrupt the stranger. While we may still say that they acted rightly, they did not act virtuously. For an act to be virtuous in the responsibilist sense, it must be based on good epistemic motives. A right act, in contrast, does not need to be based on good epistemic motives. That is not to say that a right act is not (at least in some sense) good. But a right act is not what we are concerned with when considering either AFR or the traditional view.

This example shows that, in everyday use, we focus on the acts themselves rather than on whether they issue from a virtuous trait presumably possessed by the agent. The variation of the example illustrates that a central feature of virtue responsibilism is that virtues require good epistemic motives (Zagzebski, Reference Zagzebski1996; Montmarquet, Reference Montmarquet1993; Battaly, Reference Battaly2008; Baehr, Reference Baehr2011). A virtuous person cares deeply about epistemic goods, such as having true beliefs, attaining knowledge, or gaining understanding, so that, out of this fundamental concern, the agent performs acts of epistemic attentiveness, courage, honesty, open-mindedness, fair-mindedness, carefulness, thoroughness, humility, or what have you. Epistemic responsibilist virtues are fundamentally rooted in a deep and abiding desire for knowledge and understanding. As Zagzebski eloquently puts it, they are rooted in a desire for “the motivation for truth or cognitive contact with reality” (Zagzebski, Reference Zagzebski1996: 167). Hereafter, I will refer to the required good epistemic motives as the motivation for truth.

Motivational states are act-directing reasons. They explain why a person acts in a particular way rather than another. We do not praise a person for performing acts based on bad or vicious motives. It is the motivation that a person acts upon that reveals something significant about who they are as a person. Regarding virtue responsibilism, good epistemic motives reveal what a person cares about and values. They are what make one praiseworthy.

Capturing the required good epistemic motives is central not only to the traditional account of epistemic responsibilist virtue but also to any view that takes the performance of virtuous acts as primary. For an act to count as virtuous in the responsibilist sense, it needs to be based on the motivation for truth. The motivation for truth does not single out specific internal states. Rather, it is to be understood as a subordinate notion that includes a variety of mental states or clusters of mental states, such as wishes, desires, longings, aspirations, inclinations, or what have you (this will become important in §2). These mental states or clusters of mental states occur consciously or unconsciously “at a particular moment of time or span of time” (Zagzebski, Reference Zagzebski1996: 132) and tend to be persistent.

However, as pointed out by the example of the stranger, being motivated for truth alone does not suffice to perform a virtuous act. One also needs to act in accordance with the epistemic end of the virtue in question.

Every virtue can be characterized with respect to its particular epistemic end. The epistemic end of open-mindedness can be characterized as the ability to “restrain the natural impulse to ignore or be unreceptive to the views contrary to one’s own” (Zagzebski, Reference Zagzebski1997: 10). A person who possesses the virtue of open-mindedness is “willing to consider alternative views, and if these seem more accurate or well founded, then [they are] willing to change and revise [their] initial beliefs. This is epistemically beneficial because it allows us to form the best, most reasonable and most accurate beliefs we can” (Heersmink, Reference Heersmink2017: 4). To act in accordance with the epistemic end of open-mindedness is, for example, not to ignore contrasting views or to be biased in one’s favor.

The same holds for other virtues. Here is another example: the epistemic end of the virtue of attentiveness can be characterized as exhibiting “a general alertness concerning the object of inquiry” (Baehr, Reference Baehr2011: 19). A person who possesses the virtue of attentiveness pays “close attention and focus[es] on the task at hand. An attentive person has a sustained focus when performing some cognitive task, say, writing an essay or reading a text. He or she has a sustained attention to important details, and processes these details in an adequate way” (Heersmink, Reference Heersmink2017: 4). To act in accordance with the epistemic end of attentiveness is, for example, to focus on the task at hand or to process details in an adequate way.Footnote ¹

Having clarified what an act that is in accordance with the epistemic end of the virtue in question amounts to, or, for short, what it is to “act ψ-ly” (ψ being any epistemic virtue), the definition of a performance of a virtuous act, which captures the right link between the motivation and acting ψ-ly, can be given as follows:

To state that an agent primarily needs to be motivated for truth means that the motivation for truth needs to be the most salient factor that leads the agent to perform a ψ-ly act. We want to exclude cases in which an agent has other, stronger, or even competing motivations that are salient, based on which they perform a ψ-ly act. For example, we want to exclude cases in which an agent acts ψ-ly not because they are motivated for truth, but because they are motivated by wealth, fame, appraisal, or because they do not want to interrupt. A ψ-ly act based on such motivation would simply not be the performance of a virtuous act.

With this definition of a virtuous act in place, the possession of a virtuous trait can be defined derivatively in terms of the performance of virtuous acts, recognizing that virtuous traits are domain- or situation-specific and vary in degree.

2. A new dispositional model of virtue

Let me begin by asking the following question: Why should we understand virtues as being a single disposition, when a virtue involves not only acting in accordance with the epistemic end of the virtue in question but also being motivated for truth? Consider the virtue of epistemic open-mindedness again. For an agent to possess the virtue of open-mindedness, it is not only required that they act open-mindedly, such as considering alternative views or changing or revising their initial beliefs. An open-minded person also needs to act open-mindedly because of their motivation for truth, such as desiring to believe the truth or wishing to gain understanding. Instead of conceiving virtue as a single disposition that leads the agent to perform virtuous acts, as the traditional view indicates (Aristotle NE; Hursthouse, 1999; Zagzebski, Reference Zagzebski1996, 2004; Annas, Reference Annas and Copp2005; Battaly, Reference Battaly2008; Baehr, Reference Baehr2011; Alvarez, Reference Alvarez2017), two dispositions seem to be involved. Let us, thus, conceive of a virtue as consisting of the following two dispositions (Miller, Reference Miller2013, Reference Miller2014):

A virtuous trait of ψ consists of the disposition to be motivated for truth (D1= {d1₁, d1₂, …, d1_n}; (n ∈ ℕ)) and the disposition to act ψ-ly (D2= {d2₁, d2₂, …, d2_m}; (m ∈ ℕ)).

Thus, to give an example, the virtue of open-mindedness consists of (D1) the disposition to be motivated for truth and (D2) the disposition to act open-mindedly. But instead of claiming that D1 triggers D2 directly, the view holds that the manifestations are relevant: the manifestations of D1 trigger D2, which manifests itself in an open-minded act. See the following scheme for a visual representation of the structure of the view:

When an agent possesses a virtuous trait, their performance of a virtuous act is typically informed by the chain involving the relevant dispositions D1 and D2 in the following way: the relevant dispositions D1 and D2 manifest themselves and are connected in a certain way, such that the manifestations M1 of D1 trigger D2, which then manifests itself in a ψ-ly act, M2. To go through the chain – i.e., whereby the manifestations of D1 trigger D2, which manifests in a ψ-ly act – simply is to perform a virtuous act (see Fig. 1). An agent who possesses a virtuous trait reliably goes through the chain, i.e., performs virtuous acts, in the relevant type(s) of situation.

Figure 1. The chain model.

However, note that, since a virtuous act is conceptually and evaluatively primary, an agent can perform a virtuous act even if the relevant dispositions (D1 and D2) are not present or not what lead the agent to be motivated for truth, on the basis of which they perform a ψ-ly act. Recall the example of your friend who carefully listens to the speaker arguing for cutting social welfare programs to reduce the national deficit. By carefully listening to the stranger because they want to gain understanding, the agent performs a virtuous act of open-mindedness, regardless of whether the relevant dispositions are present or play a role in their act. To act in accordance with the epistemic end of the virtue in question, based on the motivation for truth, is what generally counts for an act to be virtuous. A ψ-ly act based on the motivation for truth is virtuous, regardless of whether D1 and D2 are present or relevant for the performance.Footnote ⁵

That said, let me elaborate on the chain model in more detail. A virtuous trait of ψ consists of D1 (D1= {d1₁, d1₂, …, d1_n}; (n ∈ ℕ)) and D2 (D2= {d2₁, d2₂, …, d2_m}; (m ∈ ℕ)). D1 encompasses a variety of mental state dispositions or clusters of mental state dispositions, such as

Accordingly, D1 manifests itself in the respective mental states or clusters of mental states, such as M1 (M1= {m1₁, m1₂, …, m1_n}; (n ∈ ℕ)):

In a similar manner, D2 (the disposition to act ψ-ly) encompasses a bundle of different dispositions or clusters of dispositions, which include internal mental state dispositions, such as (d2₁) the disposition to think ψ-ly, and/or (d2₂) the disposition to form a ψ-ly belief and/or, etc. But D2 can also be an external disposition, such as (d2₃) the disposition to perform a ψ-ly action. To exemplify with the virtue of open-mindedness, the disposition to act open-mindedly can be:

Accordingly, the manifestations M2 (M2= {m2₁, m2₂, …, m2_m}; (m ∈ ℕ)) of D2 can be internal or external since the epistemic end of a virtue can manifest itself through actions, thoughts, feelings, etc. To illustrate with the virtue of open-mindedness, M2 of D2 are, for example:

The crucial point about the model presented is that the two dispositions, D1 and D2, can be possessed independently of one another, and that their possession is independent of whether an agent possesses a virtuous trait or performs a virtuous act. What matters are the manifestations of the dispositions. The agent needs to manifest the relevant dispositions, and the dispositions need to be connected in a certain way to perform a virtuous act, such that the manifestations M1 of D1 trigger D2, which then manifests itself in a ψ-ly act, M2.Footnote ⁶ What counts for whether an agent possesses the virtuous trait of ψ is that they reliably go through the chain of the relevant dispositions, i.e., they reliably perform virtuous acts of ψ in the relevant type(s) of situation. And one can go through the causal chain of the two dispositions either sequentially or simultaneously.

By understanding virtuous traits in this way, one can still take a realist position about traits. Trait dispositions, if possessed by an agent, are genuine properties or, more precisely, instantiations of genuine properties possessed by that agent. Someone can possess those properties “independently of the presence of any stimulus event and independently of our ability to recognize or conceptualize them” (Miller, Reference Miller2014: 22).

As argued in the last section, virtue is a matter of degree (Harman, Reference Harman1999; Miller, Reference Miller2013, Reference Miller2014; Alfano, 2014; Upton, Reference Upton2016; Athanassoulis, 2000). The more often an agent performs virtuous acts in the relevant type(s) of situation, the higher the degree to which they possess the virtuous trait. The primary value is conferred on the performance of virtuous acts, and the higher the degree to which the virtuous trait is possessed, the better the agent. This is already captured by Possessing a Virtuous Trait of ψ. In addition, the chain model presented above provides the metaphysical basis to explain this.

Possessing a virtue, according to the model, is not so much about possessing the relevant dispositions (although they are a significant part), but rather about altering and strengthening the connections between the relevant dispositions to such an extent that one is more likely to go through the chain of the relevant dispositions, i.e., to perform virtuous acts.Footnote ⁷

D1 and D2 can be understood as switches that, if they are manifested, can trigger another disposition. An agent who performs a virtuous act goes along a specific track in which the relevant switches – the relevant dispositions – are activated, become manifest, and trigger the relevant subsequent disposition (Wu, Reference Wu2014, Reference Wu2016). An agent who acquires a virtue strengthens or alters the connections between the relevant dispositions in certain type(s) of situations. A person who has acquired a virtuous trait has biased or strengthened the connections between the dispositions in such a way that they are more likely to be activated in certain type(s) of situations, such that a person who has acquired a virtuous trait is more likely to go along the relevant paths for the performance of a virtuous act. Put differently, the more often one goes along certain paths, the more one biases the switches, strengthening or altering the connections.

If the connections between the dispositions have been sufficiently altered or strengthened so that one is more likely to go along the relevant paths for performing a virtuous act in certain type(s) of situations, one has acquired a virtuous trait in those type(s) of situations. If the connections have been altered or strengthened to such an extent that one, no matter the type of situation, performs virtuous acts, one has acquired a virtue simpliciter.

This can be illustrated with reference to neural network views. One virtue of this model is that it fits nicely with such views (Wu, Reference Wu2014 & Reference Wu2016; Greenwald et al., Reference Greenwald, Banaji, Rudman, Farnham, Nosek and Mellott2002; Amodio & Lieberman, Reference Amodio, Lieberman and Stangor2009; Kahneman, Reference Kahneman2011; Plous, Reference Plous1993). Whether or not neural network models are correct, they capture something essential about the human mind, and we can make use of those insights.

According to neural network views, the agent’s mind can be understood as a network of nodes (see Fig. 2). These nodes are interconnected by pathways or connections (Kahneman, Reference Kahneman2011; Nickerson, Reference Nickerson1998; Plous, Reference Plous1993). D1 and D2 are understood as clusters of dispositions: D1= {d1₁, d1₂, …, d1_n} and D2= {d2₁, d2₂, …, d2_m}. In this framework, d1₁, d1₂, …, d1_n and d2₁, d2₂, …, d2_m are considered the relevant nodes. When the nodes corresponding to the dispositions representing D1 and D2 are activated, it means that the dispositions have manifested.

Figure 2. Illustration of the chain model in reference to a neural network.

Let me illustrate this with an example. Imagine someone who initially doubts the existence of climate change due to preconceived biases, leading them to favor information that aligns with their preconceptions and dismiss evidence to the contrary. At the outset, the person’s network is heavily biased, with strong connections reinforcing their skepticism about climate change and weak or non-existent connections to nodes representing D1 and D2.

As the person encounters new information – scientific studies, expert opinions, and empirical data – these inputs activate previously underutilized nodes in their network (Kwong, Reference Kwong2019). For instance, the node representing the desire to believe the truth (d1₁) is activated, which in turn leads to the activation of the node associated with considering alternative ideas (d2₁).

The processes relevant to the virtue of open-mindedness involve altering and strengthening the connections between the nodes that correspond to D1 and D2. As the person processes credible evidence, the connections between nodes representing D1 and D2 become stronger, while connections to other nodes, such as those that reinforce preexisting beliefs, misconceptions, or self-interest, weaken (Beck, Reference Beck2011; Cooksey & Freebody, Reference Cooksey and Freebody1986). This process is not instantaneous but occurs iteratively, with repeated exposure to evidence further reinforcing accurate connections and weakening erroneous ones (Clark, Reference Clark1997; Hinton & Sejnowski, Reference Hinton, Sejnowski, Rumelhart and McClelland1987; Sporns & Betzel, Reference Sporns and Betzel2016). Just as a neural network updates its internal parameters to better capture the relationships in the data it processes, the person’s mental network gradually restructures itself to reflect a more accurate understanding of reality.

Feedback mechanisms also play a crucial role in this process, as they strengthen connections between nodes corresponding to D1 and D2 that contribute to correct outputs, and diminish those that lead to errors. As the agent receives feedback, for example, from trusted peers, empirical observations, or the consequences of their actions, they refine the connections in their network (O’Doherty, Reference O’Doherty2012; Clark, Reference Clark1997).

Over time, the strengthening and reorganization of connections between the nodes that represent D1 and D2 lead the person to acquire the virtue of open-mindedness. The presented model illustrates that the acquisition of a virtue is a dynamic process of activating nodes that correspond to the clusters of dispositions that represent D1 and D2, reevaluating connections, and restructuring relationships between the relevant nodes. This, in turn, leads to a more accurate and flexible understanding of how and what it is for an agent to acquire a virtuous trait.

By taking virtuous acts as evaluatively and conceptually primary, AFR not only provides an adequate gradable account of virtue but, by conceiving virtues as consisting of two independent dispositions that need to be connected in a certain way, also explains what metaphysically underlies the possession of a virtuous trait. The view maintains that a ψ-ly act based on the motivation for truth is virtuous regardless of whether the relevant dispositions are present. When an agent possesses D1 and D2 and goes through the chain of dispositions, their performance of a virtuous act is an act from a virtuous trait. If an agent acts ψ-ly because they are motivated for truth but do not possess D1 and D2, or do not go through the chain of the relevant dispositions, they perform a virtuous act. According to the view presented, both acts are equally valuable.

Furthermore, the claim that virtue applies on two levels is preserved. On the local level, it applies to the acts that an agent performs based on good motives. On the global level, virtue applies to the traits that an agent possesses. An agent possesses a virtuous trait if they have strengthened or altered the connections between the relevant dispositions, D1 and D2, such that they are likely to perform a virtuous act in certain type(s) of situations.

Let me now turn to the objections against AFR. One might argue that, by taking virtuous acts as evaluatively and conceptually primary, AFR is, first, normatively slight; second, that it fails to meet the demands of an exemplarist virtue theory; and third, that it does not satisfy all the predictive and explanatory requirements of a virtue responsibility theory.

3. AFR is not normatively slight and does not fail to meet the demands of an exemplarist virtue theory

Virtue epistemology is a normative discipline. The normativity of virtue epistemology is tied to the virtuous trait: we ought to act and behave like the (fully) virtuous agent. Only the agent who is fully virtuous – that is, who possesses the virtuous trait – is praiseworthy.

That normativity is tied to virtuous traits becomes most evident in reference to the situationist challenge, which threatens the very existence of virtuous traits. The situationist challenge can be formulated as a dilemma: on the one hand, if virtues are understood globally, such as conscientiousness-no-matter-the-type-of-situation, then research in social psychology suggests that they are empirically unfounded. On the other hand, if virtues are understood locally, like conscientiousness-in-circumstances-C, then virtues appear to lose their normative power (Olin and Doris, Reference Olin and Doris2014: 679f.; Sylvan, Reference Sylvan, Fairweather and Alfano2017: 139). As Abrol Fairweather (2017) puts it:

The traditional view conceives virtue as person-attaching. On this view, virtues are thick evaluative properties, such that we call a person conscientious, courageous, or what have you. We are concerned with the evaluation of agents’ praiseworthiness, according to the traditional view. The normativity is then tied to the virtuous traits that an agent possesses. To state that a person is conscientious, courageous, etc., and to praise them as the person they are would indeed be strange if they only perform virtuous acts in this one specific domain or with regard to this one particular subject matter. We would not call a person conscientious or attentive if the trait led the agent to perform virtuous acts only in a very specific type or types of situations. Traits such as conscientious-when-it-comes-to-making-coffee or attentive-while-watching-the-news-on-TV do not confer the praiseworthiness associated with virtues, or so the objection goes.

Many have noted (Harman, Reference Harman1999; Doris, Reference Doris2002; Alfano, Reference Alfano2013; Hurka, Reference Hurka2006; Sylvan, Reference Sylvan, Fairweather and Alfano2017) that a view which focuses on acts rather than virtuous traits does not face the situationist challenge, and so neither does AFR. If a person acts open-mindedly in situation S₁ but not in a trivially different situation S₂, then situation S₁ promotes open-minded acts while situation S₂ does not (Hurka Reference Hurka2006: 75). That people are influenced by trivially seeming features of the situation will have no effect on the concept of virtuous acts, which is just that of ψ-ly acts from good epistemic motives, regardless of whether they issued from a virtuous trait. An act that is in accordance with the epistemic end of the virtue in question and based on good epistemic motives is virtuous, full stop. However, similar to the situationist challenge’s critique of local traits, one could argue against AFR by claiming that act-attaching virtue properties lack substantial normative significance; that, in other words, by advocating that individual acts based on good epistemic motives are virtuous, responsibilism loses its normative appeal.

But what exactly do we mean by normatively slight? If normatively slight just means attaching to acts, this hardly seems problematic. Certainly, there is a sense in which taking acts based on good epistemic motives as virtuous is slight; after all, we are just concerned with individual performances of virtuous acts. However, this does not diminish their normative appeal (Sylvan, Reference Sylvan, Fairweather and Alfano2017: 152). Why do we care about epistemic virtues in the first place? We care about virtues in the classic responsibilist sense because they make it probable that one performs virtuous acts. What ultimately counts are the virtuous acts, even for a traditional account. As Sylvan notes, “[t]his pattern of explanation is hardly uncommon: we care about the disposition to believe the truth because we care about true belief, we care about the disposition of happiness because we care about happiness, and, similarly, we care about the disposition to do virtuous things because we care about virtuous things” (2017: 152).

We should, therefore, not be dissatisfied with a theory that takes acts based on good epistemic motives to be virtuous, irrespective of whether they issue from a virtuous trait. AFR does everything that we want a virtue responsibilist theory to do and more. The only difference is that AFR denies that “[t]he relevant thick virtue properties are character traits” (Sylvan, Reference Sylvan, Fairweather and Alfano2017: 141). Rather, “[t]he relevant thick virtue properties are act-attaching ones or ones analysable in terms of these (e.g., local person-attaching properties)” (Sylvan, Reference Sylvan, Fairweather and Alfano2017: 141).

A person who (reliably) performs virtuous acts and thus possesses a virtuous trait is still praiseworthy and admirable even if they possess the trait only locally. An agent who possesses a local trait can still serve as someone worth imitating. We praise and admire the virtuous trait of thoroughness possessed by a physicist, for example, when they conduct their experiments and consider their acts worth imitating, even if they do not perform virtuous acts of thoroughness in any other domain. We praise and admire the virtuous trait of attentiveness possessed by the documentary filmmaker David Attenborough when it comes to attentively observing and representing nature and wildlife, and consider his virtuous acts of attentiveness worth imitating, even if he might not perform such acts with respect to other domains.

Not only do we praise and admire more-or-less local virtuous traits that an agent possesses, but even the single performance of a virtuous act can be so praiseworthy and admirable that it is taken as exemplary and worth imitating. For example, when Martin Luther King Jr. gave his first speech after the rectory where he lived with his family was bombed in 1956, in which he defended his view against White supremacy, we consider this virtuous act of courage to be so praiseworthy and admirable that it is worth imitating.

One might object that King is one of the paradigm examples of a courageous agent. But in fact, we ascribe the virtuous trait of courage to him precisely because of the many various virtuous acts he performed. We consider defending one’s view if one’s life is in danger to be admirable and praiseworthy, even if the agent only did so once. And in fact, we consider such an act admirable, praiseworthy, and worth imitating, irrespective of whether the act issued from a virtuous trait. That is precisely what AFR claims.

You are not convinced? Consider another example. When Colin Kaepernick first ‘took the knee’ on September 1, 2016, to protest and raise attention about racial inequality and police brutality in the United States, his virtuous act of courage was so praiseworthy and admirable that it was considered worth imitating. As a matter of fact, taking the knee in sports is nowadays common practice to raise attention against racism.

Here is another example. When Thomas Egerton, responding to widespread prejudice against female authors, purchased the rights to Jane Austen’s Pride and Prejudice in the 18^th century, we consider his act of open-mindedness to be praiseworthy, admirable, and worth imitating, even though he did not perform virtuous acts of open-mindedness regarding female authorship in other situations. It is commonly known that he deprived Jane Austin of the profits he made by selling her book.

Note that AFR does not deny that an agent who possesses a virtue, i.e., one who reliably performs virtuous acts in the relevant type(s) of situation, is the better epistemic agent (see §2). The primary value attaches to virtuous acts. However, the more often an agent performs virtuous acts in different type(s) of situations, the better an epistemic agent they become. Instead of claiming that normativity is person-attaching, such that “You ought to act as the agent who possesses virtue,” AFR states that normativity is act-attaching: “You ought to perform virtuous acts.”Footnote ⁸ And the more often you perform virtuous acts, the more virtuous you are as an agent.

At first glance, AFR may seem to contrast with the exemplarist tradition (Zagzebski Reference Zagzebski2017, Reference Zagzebski2010), which seeks a framework that is practical, emotionally resonant, and an accessible way of understanding and cultivating virtue. One might worry that a theory which takes virtuous acts as conceptually and evaluatively primary may not be able to meet these demands.Footnote ⁹

According to Linda Zagzebski’s exemplarist virtue theoryFootnote ¹⁰ , our understanding of epistemic excellence begins with admiration for epistemic exemplars – individuals whose actions and character embody epistemic virtues such as curiosity, humility, or open-mindedness in exemplary ways. Rather than deriving epistemic virtues from abstract principles, Zagzebski argues that these virtues are recognized through the traits admired in exemplary thinkers. Intellectual growth involves emulating such figures, thereby fostering the development of an intellectually virtuous character. The approach places admiration and emotional engagement with virtuous agents at the foundation of epistemic reflection, providing a practical, narrative-based, and relational framework for cultivating intellectual virtues. Thus, at the core of exemplarist virtue theory are the agents who possess virtuous traits. This seems to conflict with AFR.

However, even Zagzebski acknowledges that an exemplarist theory does not have to center (solely) on a person and their virtuous traits, but may also focus on exemplary acts (Zagzebski Reference Zagzebski2010: 54). Like virtuous persons, exemplary acts can be defined by direct reference: a good person is a person like that, and we can likewise say that a good act is an act like that (Kripke Reference Kripke1980, Putnam Reference Putnam1979). In principle, both good persons and good acts can serve as the foundation of an exemplarist theory without the use of descriptive concepts (Zagzebski Reference Zagzebski2017: 51).Footnote ¹¹ Nonetheless, as Zagzebski points out, the concern is that “reference to a few exemplary acts will not be helpful in constructing a comprehensive [exemplarist] theory.”

As illustrated by the examples of the physicist, David Attenborough, Martin Luther King Jr., Colin Kaepernick, and Thomas Egerton, we at least sometimes attach normativity directly to acts. However, this does not imply that the traits which an agent possesses are not of normative significance and cannot serve, alongside virtuous acts, as the foundation of an exemplarist theory. Virtuous acts are taken as conceptually and evaluatively primary, and the possession of a virtuous trait is defined in terms of the performance of virtuous acts. Virtuous traits come in degrees: the more often an agent performs virtuous acts in the relevant type(s) of situations, the higher the degree to which the virtuous trait is possessed. The higher the degree to which the trait is possessed, the more excellent the agent becomes. Thus, AFR allows that agents serve as exemplars precisely because they do and would perform virtuous acts in the relevant type(s) of situation.

Recall that virtue applies on two levels. On the global level, it applies to the traits that an agent possesses. On the local level, it applies to the acts that an agent performs based on good motives. According to the traditional view, virtuous traits are conceptually and evaluatively primary: a good person is a person like that, and virtuous acts are defined derivatively as acts which a virtuous agent would perform. In contrast, AFR takes virtuous acts as both evaluatively and conceptually primary and holds that a good person is a person who performs acts like that. The more often an agent performs virtuous acts, the more excellent and exemplary the agent. The normative criterion is thus formulated as follows: “You ought to perform virtuous acts” (plural!), and the more often you perform virtuous acts, the more virtuous and the more excellent you are as an agent.

In fact, AFR not only provides a more comprehensive account of how and why people want to acquire virtuous traits, but it also satisfies all the predictive and explanatory demands of a virtue responsibilist theory.

4. Virtues help us to explain, predict, evaluate, and understand Agents

Besides the demand of not being normatively slight, any virtue responsibilist theory also needs to be predictive and explanatory. Virtues are “‘thick’ in the sense that they are descriptively rich (knowing that someone is [courageous] tells you a lot about [them]), explanatory powerful (the [courageous] person not only does but would [act courageously] in circumstances where the same reasons were in force) and evaluatively laden ([courage] is admirable) [and worth imitating]” (Alfano, Reference Alfano2012: 229f., my emphasis).

By understanding virtues as “thick,” “it ought to be possible to use them not only for evaluation of agents’ praiseworthiness but also for predicting and explaining how agents inquire and believe” (Alfano, Reference Alfano2012: 234). After all, why should we care about epistemic virtues if they fail to be of explanatory and predictive significance?

By taking virtuous acts as conceptually and evaluatively primary, all predictive and explanatory demands of a virtue responsibilist theory are satisfied. We are able to explain, predict, understand, and evaluate human behavior. Although AFR conceives of virtue as primarily act-attaching and defines person-attaching virtue in terms of the former, it preserves the following two essential features of a virtue theory (Miller, Reference Miller2014: 380):

1. (i) A theory of virtues that accounts for development and education concerning the kind of character that agents want to cultivate. In contrast to the traditional account, AFR does not focus on a rarefied ideal, but instead on traits that agents can actually develop.

2. (ii) An account that supports our ordinary practice of appealing to virtues in explaining human behavior.

Regarding (i), it does not follow from a conception of epistemic responsibilist virtues that takes virtuous acts as basic that one has no grounds to explain how one acquired a virtuous trait. Virtuous traits can still be acquired through gradual habituation, or a process of repetition over a considerable period of time (Zagzebski, Reference Zagzebski1996: 117). Acquiring an intellectually virtuous trait still develops in stages and “begins with the imitation of virtuous persons, requires practice which develops certain habits of feeling and acting” (Zagzebski, Reference Zagzebski1996: 150).

People can still be brought to the ideal of possessing an epistemically virtuous trait and regard those who possess virtuous traits, no matter how local, as individuals worth imitating. As demonstrated by the examples of David Attenborough, the physicist, Martin Luther King Jr., Colin Kaepernick, and Thomas Egerton, we can still praise and admire the most local traits, and even the individual performance of virtuous acts, and take them to be worth imitating. Further, we can still imagine a virtuous agent and the acts they would perform. We can still take such an ideally virtuous agent as an exemplar and try to imitate them. Admittedly, there is no guarantee that one can ever achieve the goal of acquiring a virtuous trait, but this does not mean that the account offered cannot provide an answer to how and why people acquire and, in fact, want to acquire virtuous traits and perform virtuous acts.

AFR provides an act-centered form of exemplarism. In addition to the admiration of traits which an agent possesses, the admiration of single performances of virtuous acts plays a key role not only in what counts as a virtuous act but also in recognizing who possesses a virtuous trait. Through admiration, we cultivate performances of virtuous acts and, consequently, virtuous traits by emulating the actions of exemplars.

A virtue theory that focuses on acts rather than on the virtuous traits which an agent possesses is still grounded in narrative practices, as well as relational and emotional aspects (Zagzebski, Reference Zagzebski2010 & Reference Zagzebski2017). AFR emphasizes learning from real-world examples rather than abstract theory. It promotes narrative learning, whereby we engage with the stories of exemplary thinkers who perform virtuous acts. Through these narratives, we become emotionally invested in the significance of performances of virtuous acts and develop a desire to internalize and perform them ourselves. However, in contrast to Zagzebski, AFR allows for a broader spectrum of cases. It involves not only admiring and emulating individuals who possess virtuous traits but also admiring and emulating individual performances of virtuous acts.

As a matter of fact, AFR does even better in explaining why one should want to acquire a virtuous trait or perform virtuous acts. According to the traditional view, possessing a virtuous trait is an ideal that few will ever be able to attain (Code, Reference Code1984; Olin and Doris, Reference Olin and Doris2014; Upton, Reference Upton2016). The development of virtue is influenced by various factors that are often a matter of chance. Achieving virtue requires numerous favorable conditions in one’s life, including the influences one experiences and the situations one faces (Aristotle NE, 1100a5–9 and 1100a5–7; Athanassoulis, Reference Athanassoulis, Church and Hartman2019: 21).

The relevant question one should ask regarding the dominant traditional view’s conception of virtue is the following: Why should one even try to acquire a virtuous trait if its development depends on features largely subject to luck? Ultimately, according to the traditional view, it remains for the “fortunate, talented, and diligent” few (Olin and Doris, Reference Olin and Doris2014: 677) to acquire virtues, leaving the majority unlikely to become virtuous agents (Upton, Reference Upton2016).Footnote ¹²

Don’t we want a conception of virtue that allows agents to be virtuous, even if that means only once? Especially in times of climate change and economic crises, we need a conception of virtue that enables individuals to become better epistemic agents, allowing them to acquire a virtuous trait or to become virtuous agents with respect to those situations or topics they are interested in and care about. For educational purposes, it is better to educate people to acquire virtuous traits that they are actually able to realize. We need a conception of virtue that allows us to say, “The act you performed based on good epistemic motives is virtuous” instead of “Your act is merely one a virtuous agent would perform, but unlucky you, is not virtuous,” as proponents of the traditional view have to say.

AFR can account for all of these demands. By attaching virtue to acts and constructing virtuous traits out of the former, the view allows people to actually acquire virtuous traits and not only to chase an unreachable ideal. According to AFR, performing virtuous acts is not a rare ideal reserved, if at all, for the fortunate, diligent, and sufficiently talented. Everyone can perform a virtuous act and acquire even the most local traits. As long as the acts are in accordance with the epistemic end of the virtue in question and based on good epistemic motives, they will be virtuous, no matter whether the agent possesses an underlying virtuous trait. A virtue responsibilist theory should focus on and emphasize the development and education of performances of virtuous acts and traits which an agent is able to realize (Miller, Reference Miller2014; Code, Reference Code1984). The goal should not be to possess a virtuous trait in order to perform virtuous acts, but to become a better epistemic agent by reliably performing virtuous acts in more type(s) of situations.

Note again that, by stating that virtue attaches to acts, this is not to say that AFR cannot account for the existence of better epistemic agents. Clearly, someone who possesses a virtuous trait to a higher degree is the better epistemic agent. The new dispositional model provides the metaphysical foundation for this (see §2). On the global level, virtue applies to the traits that an agent possesses. The agent possesses a virtuous trait (metaphysically speaking) if they have strengthened and altered the connections between the relevant dispositions D1 and D2, such that they reliably perform virtuous acts in more type(s) of situations. The more they strengthen and alter the connections between the dispositions, the more often they perform virtuous acts in the relevant type(s) of situations, and thus the higher the degree to which they possess the virtuous trait. On the local level, virtue applies to the acts that an agent performs based on good motives. It is the latter that should be taken as conceptually and evaluatively primary. What AFR denies is that an act which an agent performs based on good epistemic motives is not virtuous if it did not issue from a virtuous trait. Put differently, an act that is in accordance with the epistemic end of the virtue and based on good epistemic motives is virtuous, full stop. This, however, does not imply that an agent who possesses a virtuous trait to a higher degree, i.e., an agent who more often performs virtuous acts in the relevant type(s) of situation, is not the better epistemic agent.

Concerning feature (ii), it does not follow from an account that allows for narrow virtuous traits that one cannot turn to our ordinary practice of appealing to virtuous traits to explain behavior. Just because most of us only possess local virtuous traits, this does not imply that virtuous traits are useless for providing causal explanations. Even though the view offered allows for local virtuous traits, and people might sometimes fail to perform virtuous acts in certain type(s) of situations or while being confronted with certain topics, it does not follow that trait attribution is genuinely false or that virtuous traits are not an important feature that help us explain why people act the way they do in certain situations.

AFR not only permits that one can explain the behavior of a person through the possession of more or less local virtuous traits but also vindicates the consistency of those traits: an agent who possesses a local trait does and would perform virtuous acts in certain type(s) of situations. Moreover, since AFR analyses the possession of a virtuous trait in terms of the performance of virtuous acts, and the performance of an epistemically virtuous act always concerns a certain situation, situations are of crucial importance right from the start.

We have ample reasons to put act-attaching virtues first. We can still use virtue terms to understand a person better. Knowing that a person possesses a local virtuous trait with respect to particular type(s) of situations still helps us to understand who they are as a person. Reference to even the most local trait a person possesses helps us to explain why someone acts a certain way. If an agent possesses a virtuous trait with respect to situations of type S_{1, …, n}, then they do and would perform virtuous acts in (similar) situations of type S_{1, …, n}. Thus, we can use virtuous traits to predict how someone is going to behave, such as what they are most likely to think and feel now and in the future.