 |
| Ernest Sosa |
Ernest Sosa is Romeo Elton Professor of Natural Theology and Professor of Philosophy at Brown University and regular Distinguished Visiting Professor at Rutgers University. He received his Ph.D. from the University of Pittsburgh and has taught at Brown since 1964. Since 1983, he has been the editor of Philosophy and Phenomenological Research and since 1999, with Jaegwon Kim, the co-editor of Nous.
Sosa has published essays on issues in a wide variety of philosophical areas such as metaphysics, logic, philosophy of mind, theory of action, and philosophy of language, but he has been most influential in epistemology, where he is known for advocating a virtue-based approach to the analysis of knowledge and justification with an emphasis on the importance of a reflective perspective.
What is distinctive of virtue epistemology is the order of explanation: A belief ’s epistemic status is to be understood in terms of the epistemic properties of the subject, which in turn are to be captured by employing the concept of an intellectual virtue. How is this concept to be understood? In pure virtue epistemology, construed in analogy to pure virtue ethics, the concept of an intellectual virtue is basic (Foley 1994).
 |
| Jaegwon Kim |
Sosa, however, conceives of an intellectual virtue as a stable disposition to form true beliefs in a certain field of propositions, F, under suitable circumstances, C. Thus his brand of virtue epistemology, which he has labeled virtue perspectivism, is not an example of the pure kind but may be viewed as a form of reliabilism.
The two main elements of Sosa’s virtue perspectivism are the concepts of an intellectual virtue and an epistemic perspective. As already indicated, Sosa conceives of intellectual virtues in terms of reliability. Reliably functioning faculties, such as vision, hearing, introspection, and memory are examples of intellectual virtues. Sosa calls beliefs that are grounded in the exercise of such virtues apt.
Apt beliefs, if true, qualify as knowledge, or, more precisely, as animal knowledge, to be distinguished from reflective knowledge. With the distinction between these two kinds of knowledge, the second main element of virtue perspectivism comes to the fore: the concept of an epistemic perspective.
 |
| epistemic perspective |
Let S refer to the subject whose beliefs we wish to evaluate. Suppose S’s visual belief that p is true and, due to the reliability of S’s vision, apt. Hence by employing her faculty of vision, S acquires animal knowledge that p.
For S’s belief to rise to the level of reflective knowledge, a further condition must be met: S must form a meta-belief to the effect that her belief and its being true have their origin in a reliable faculty.
In general terms, if from S’s epistemic perspective, a faculty is coherently viewed as reliable within field F and circumstances C, then by employing this faculty S can acquire reflective knowledge within field F and circumstances C.
 |
| reflective knowledge |
Animal knowledge, then, results from external aptness: the exercise of faculties that are in fact reliable. Reflective knowledge also requires aptness, but, in addition, the adoption of an internally coherent perspective with respect to the reliability of one’s faculties. Sosa’s virtue perspectivism, then, combines both an externalist and an internalist element.
In the large body of work in which Sosa articulates and defends his approach to the philosophical explanation of knowledge and justification, he has addressed various problems that arise for virtue perspectivism.
First, there is the problem of what a reliabilist should say about what are referred to as evil demon victims: subjects whose beliefs seem justified although, due to the massive deception to which the victims are subjected, their beliefs are grounded in unreliable faculties.
 |
| virtue perspectivism |
Sosa responds that, whereas the demon victims’ beliefs are actual world justified (as the victims’ faculties would be reliable in the actual world, they are same world unjustified because the faculties the victims employ are unreliable in their own world (1994a). In more recent terminology, Sosa classifies the victims’ beliefs as adroit though not apt (Sosa 2003).
Second, Sosa’s reliability-grounded virtue perspectivism is challenged by what Sosa calls the problem of meta-incoherence, which arises from cases in which a subject’s beliefs are produced by a faculty whose de facto reliability is not (or at least not yet) recognized by the subject.
Since such subjects do not meet the perspectival condition of having formed reliability-attributing metabeliefs about the relevant belief sources, Sosa judges that the beliefs in question are unjustified, or not reflectively justified (Sosa 1991).
 |
| actual world |
Third, there is the generality problem, which for Sosa amounts to the challenge of finding the right level of specificity in describing field and circumstances. Here, Sosa’s solution is to require that the relevant descriptions be useful within the subject’s epistemic community and to the subject herself (Sosa 1991).
Three further, important problems to which Sosa has articulated detailed solutions are the following: First, how can we distinguish between accidental and non-accidental reliability? Second, what justifies reliability-attributing perspectival meta-beliefs (Sosa 1994a)?
Third, why is the process by which reliability-attributing meta-beliefs are formed (using, for example, perception to attest to the reliability of our perceptual faculties) not viciously circular (1994b and 1997)?
 |
| requires sensitivity |
Recently, Sosa has also contributed important work on the following question: If a belief is to be an instance of knowledge, what modal link must there exist between the belief and its truth? According to some, knowledge requires sensitivity: S would not believe that p if p were false.
Viewing this condition as too demanding, Sosa objects to it on the basis of the following case: Having dropped a trash bag in the garbage shoot, you believe the bag will momentarily reach its destination in the basement.
 |
| Sosa proposes safety |
This belief, Sosa suggests, amounts to knowledge even though it is not sensitive: if p (the bag will land momentarily) were false (because, say, the bag snagged in the shoot), you would still believe p. As an alternative, Sosa proposes safety: If S were to believe p, p would be true (or: Not easily would S believe incorrectly in believing that p).
Though your belief that the bag will land momentarily is not sensitive, it is indeed safe, for possible worlds in which S believes that the bag will shortly arrive downstairs, but believes this mistakenly, are indeed remote (Sosa 1999).
 |
| puzzle of skepticism |
The distinction between safety and sensitivity assumes particular significance for Sosa, for he appeals to it for the purpose of rejecting the contextualist solution to the puzzle of skepticism. Contextualists have argued that, when confronted with a skeptical argument, we face a paradox because, although we find the premises plausible, we wish to reject the conclusion.
According to the contextualist response, the puzzle is to be solved by appeal to the context-sensitivity of the word know. Sosa suggests an alternative solution: Skeptical arguments may (misleadingly) seem cogent because we fail to recognize that knowledge requires not sensitivity, but merely safety (Sosa 1999 and 2003).
 |
| Skeptical arguments |