Subjunctivitis

Abstract

Subjunctivitis is the doctrine that what is distinctive about knowledge is essential modal in character, and thus is captured by certain subjunctive conditionals. One principal formulation of subjunctivism invokes a ``sensitivity condition'' (Nozick, De Rose), the other invokes a ``safety condition'' (Sosa). It is shown in detail how defects in the sensitivity condition generate unwanted results, and that the virtues of that condition are merely apparent. The safety condition is untenable also, because it is too easily satisfied. A powerful motivation for adopting subjunctivism would be that it provides a solution to the problem of misleading evidence, but in fact, it does not.

Appendix

De Rose on the Closure Principle

The hallmark of De Rose’s view of knowledge–its superiority to Nozick’s original account–is supposed to be that it reconciles the tracking requirement with preservation of the Closure Principle. But the point is not at all straightforward. Consider the following case:

Daily Planet Case. Perry believes, falsely, that Jimmy Olson is Superman. Perry is sitting in his office looking at Clark Kent and Jimmy. Perry knows (C), that Clark Kent is in the office. Perry presumably knows the disjunction `Either Clark Kent is in the office or a native of the planet Krypton is' (C or K).

There’s no trouble about Perry’s knowing C. In the closest possible world W in which Clark is not in the office, Perry would not believe he was. So Perry satisfies D with respect to C. But Perry seems not to know the disjunction (C or K). Specifically, the standards for knowing (C or K) are set by Perry’s knowledge that C. In order to know (C or K), Perry must not believe (C or K) in the nearest -C world. The nearest -C world is the closest possible world in which Clark Kent is not in the office. But that is a world in which there is not a native of Krypton in the office either. In other words, the nearest -C world is just W, and, in W, Perry falsely believes (C or K). So, Perry’s belief that (C or K) fails to satisfy D. De Rose seems left with the result that Perry knows C, but does not know (C or K).

The only way I can see to avoid this outcome is to modify De Rose’s conditions so that they explicitly take into account the “methods” by which beliefs are formed, where “methods” are construed narrowly (see above). Thus, S knows P by method M only if

(D*) There is some proposition Z, selected by the context of knowledge attribution, such that S's belief that P matches the fact as to whether P is true in the sphere of nearby worlds in which S uses method M and which is bounded by the nearest not-Z world.

We can see how this modification might ward off failure of the Closure Principle in the Daily Planet Case. Let us say that Perry knows C and that -C sets the boundary to S in this instance. What is crucial is that Perry's actual belief that (C or K) is due to his seeing Clark in the office; his seeing Clark is part of the method M by which he actually arrives at his belief that (C or K). Perry will fail to know (C or K) only if he wrongly believes (C or K) by method M in the nearest -C world. The nearest -C world is W. In W, Perry believes (C or K) by inferring it from his beliefs that Jimmy is Superman and that Jimmy is in the office. W is not a world in which Perry employs method M; a fortiori, it is not a world in which Perry wrongly believes (C or K) by method M. Hence, D* allows that Perry knows (C or K) as well as K, and the Closure Principle is preserved. But there may be a very significant cost. De Rose's essentially reliabilist commitments lead him to downplay the contribution of evidential considerations to knowledge. We now see that this contribution is substantial and ineliminable, which calls into question the soundness of De Rose's overall approach to knowledge.