Loading field map
For much of the twentieth century, analytic epistemology focused on the properties of individual beliefs. The central question was: what turns a true belief into knowledge? The standard answer—justified true belief—collapsed in 1963 when Edmund Gettier published a series of counterexamples. In a Gettier case, a person holds a true belief that is justified, yet the truth is a matter of luck rather than cognitive achievement. A driver sees a barn-shaped façade and forms the belief "that is a barn"; unbeknownst to her, the area is full of barn façades, but she happens to be looking at the one real barn. Her belief is true and justified, but it does not seem to be knowledge.
Virtue epistemology emerged in the 1980s as a distinctive response to this pressure. Instead of analyzing belief-by-belief, it shifted the unit of evaluation to the knower. Knowledge, on this approach, is a cognitive success that is attributable to the knower's intellectual virtues—stable features of the agent that reliably produce true beliefs or manifest good intellectual character. The subfield has since grown into three major frameworks, each offering a different account of what those virtues are and how they connect to knowledge.
Ernest Sosa launched virtue epistemology in 1980 with a paper that reframed intellectual virtues as reliable cognitive faculties. On this view, a virtue is any stable disposition—such as good eyesight, memory, or reasoning—that tends to produce true beliefs rather than false ones. The knower does not need to be aware of her virtues or to have chosen them; they are features of her cognitive architecture. Sosa called this an externalist approach because the status of a belief as knowledge depends on facts about the world (the actual reliability of the faculty) rather than on the believer's internal perspective.
Sosa drew a further distinction that remains influential. Animal knowledge is the kind a creature has when its true belief is produced by a reliable faculty, even if the creature has no reflective grasp of why the belief is true. Reflective knowledge requires a second-order perspective: the knower not only has a true belief from a reliable faculty but also appreciates that the faculty is reliable. This layered picture allowed Sosa to handle Gettier cases in a new way. In the barn-façade scenario, the driver's visual faculty is not reliable in that environment, so the belief does not count as animal knowledge. The virtue condition—that the belief must be produced by a virtue—fails, even though the belief happens to be true.
Virtue reliabilism was the first systematic virtue-theoretic framework in epistemology, and it remains a leading position. Its core commitment is that knowledge is a kind of cognitive achievement: the knower succeeds because of her competence, not because of luck. Later frameworks would preserve this achievement idea while questioning whether reliable faculties are the right model for intellectual virtue.
Linda Zagzebski challenged the reliabilist picture in the early 1990s by arguing that intellectual virtues are not mere faculties but cultivated character traits. On her account, a virtue is a deep-seated disposition to think and inquire in ways that reflect intellectual motivation—a desire for truth, open-mindedness, intellectual courage, and carefulness. The analogy is with moral virtue: just as a generous person is not merely someone who reliably gives money but someone who is motivated by care for others, an intellectually virtuous person is not merely someone who reliably forms true beliefs but someone who is motivated by love of truth and exercises good judgment.
This shift from faculties to character traits had several consequences. First, responsibilism made the knower's effort and agency central. A person who happens to have good eyesight has not thereby earned any epistemic credit; credit accrues only when the knower actively cultivates and exercises virtues that require effort, such as intellectual humility or thoroughness. Second, responsibilism introduced a motivational condition: a belief counts as knowledge only if it arises from the agent's intellectual character, where character involves not just reliability but the right kind of motivation. Third, the framework opened virtue epistemology to questions about intellectual character formation, education, and the social dimensions of knowing.
Virtue responsibilism did not replace virtue reliabilism. Instead, the two frameworks entered into a productive tension that continues today. Reliabilists argued that responsibilism's motivational condition was too demanding—it would rule out much of what we ordinarily count as knowledge, especially perceptual knowledge. Responsibilists replied that reliabilism's faculty-based virtues were too thin to capture what is epistemically valuable about human inquiry. The disagreement is not about whether virtues matter but about what kind of thing a virtue is: a reliable mechanism or an acquired character trait.
By the early 2000s, Duncan Pritchard identified a problem that affected both existing frameworks. Even when a belief is produced by a reliable faculty or by a virtuous character, it can still be lucky in a way that undermines knowledge. Consider a person who forms a true belief by guessing, but whose guess is based on a reliable but unreliable-seeming hunch. The belief might be true and even produced by a reliable process, yet it seems too lucky to count as knowledge. Pritchard argued that the virtue condition alone cannot eliminate all forms of epistemic luck.
Pritchard's anti-luck virtue epistemology preserved the core insight that knowledge requires cognitive achievement attributable to the agent, but added a second condition: the belief must also be safe. A belief is safe if, in nearby possible worlds where the person forms the belief in the same way, the belief remains true. In the barn-façade case, the driver's belief is not safe because in nearby worlds where she looks at a different barn-shaped object, she would form a false belief. The safety condition captures the idea that knowledge cannot be fragile—it must be robust across slightly different circumstances.
This framework refined rather than replaced virtue reliabilism. Pritchard accepted Sosa's basic picture of knowledge as success from ability, but argued that the ability condition alone is insufficient. The safety condition is a modal constraint that works alongside the virtue condition. Anti-luck virtue epistemology thus represents a narrowing and strengthening of the reliabilist program: it preserves the externalist, faculty-based conception of virtue while adding a further anti-luck requirement. The framework has been especially influential in debates about epistemic luck and the value of knowledge.
All three frameworks remain active in contemporary epistemology, and their division of labor reflects their different strengths. Virtue reliabilism continues to be the dominant approach in mainstream analytic epistemology, particularly in work on perception, memory, and testimony. Its externalist, faculty-based model fits naturally with naturalistic approaches that draw on cognitive science. Virtue responsibilism has found a home in social epistemology and in the study of epistemic injustice, where questions about intellectual character, motivation, and power are central. The responsibilist emphasis on cultivated traits has also influenced work on education and intellectual virtue formation. Anti-luck virtue epistemology has become a leading framework in the literature on epistemic luck and the value of knowledge, and its safety condition is widely discussed even by philosophers who do not adopt the full framework.
On what do these frameworks agree? All three hold that knowledge is a kind of cognitive success that is creditable to the agent's intellectual virtues. All three reject the idea that knowledge can be analyzed solely in terms of belief properties like justification or evidence. And all three treat the Gettier problem as a symptom of a deeper issue: the need to connect knowledge to the agent's cognitive agency.
The main disagreements concern the nature of intellectual virtue and the role of luck. Reliabilists and responsibilists disagree about whether virtues are faculties or character traits, and about whether motivation matters for knowledge. Anti-luck theorists disagree with both about whether the virtue condition is sufficient or whether a separate safety condition is needed. These are not settled debates; they are live disagreements that drive ongoing research. A student entering the field today will find a subfield unified by its agent-centered approach but animated by real disputes about what it means to be a virtuous knower.