Counterfactual Epistemology: In defence of sensitivity
The claim that knowledge is subject to a sensitivity condition is widely rejected. In my PhD thesis I argue that this is a mistake: knowledge does require sensitivity, i.e. knowing p requires that one would not believe p (in the same way) if it were false.
The most influential objection to the sensitivity condition is that it is incompatible with the universal closure of knowledge under deduction. However, I argue that we have excellent reasons to reject this particular closure principle. Moreover, the sensitivity condition is uniquely well-placed to account for a restricted closure principle, which avoids the problematic instances of deductive closure while vindicating the legitimate ones. Hence this provides an argument for sensitivity.
Other objections to the sensitivity condition take the form of straightforward counterexamples. In response to the claim that sensitivity is incompatible with higher-order knowledge, I show that this argument fails to take into account the basis-relativity of sensitivity, but once we do take this into account we can see that higher-order beliefs are sensitive.
In response to the claim that sensitivity is incompatible with inductive knowledge, I show that so long as the counterfactuals relevant to assessing sensitivity are understood as evaluable via backtracking, inductive knowledge is sensitive. (And moreover, there is nothing suspicious about backtracking counterfactuals.)
In response to the claim that sensitivity is incompatible with knowledge of merely highly likely events, I argue that we do not know those things. (And if quantum mechanics implies that the entire universe functions like an enormous lottery, then quantum mechanics implies that we know less than we ordinarily think.)
After defending and arguing for the standard sensitivity condition, I go on to argue for an even stronger form of this condition. By defending the viability of simple causal or counterfactual accounts of the epistemic basing relation, in combination with the claim that believing a proposition often amounts to believing many more fine-grained things, I argue that knowing that p requires that for every q such that one believes p partly by believing q, if q were false then one wouldn’t believe q (on the same basis as one actually does). This condition is able to explain a wide range of cases, and solves the well-known generality problem for process reliabilism.
Finally, I discuss how the sensitivity condition interacts with other popular positions in epistemology (safety, virtue epistemology, hinge epistemology, and contextualism). The sensitivity condition is compatible with all of these positions and makes valuable contributions to our understanding of epistemic luck and contextualism.
The most influential objection to the sensitivity condition is that it is incompatible with the universal closure of knowledge under deduction. However, I argue that we have excellent reasons to reject this particular closure principle. Moreover, the sensitivity condition is uniquely well-placed to account for a restricted closure principle, which avoids the problematic instances of deductive closure while vindicating the legitimate ones. Hence this provides an argument for sensitivity.
Other objections to the sensitivity condition take the form of straightforward counterexamples. In response to the claim that sensitivity is incompatible with higher-order knowledge, I show that this argument fails to take into account the basis-relativity of sensitivity, but once we do take this into account we can see that higher-order beliefs are sensitive.
In response to the claim that sensitivity is incompatible with inductive knowledge, I show that so long as the counterfactuals relevant to assessing sensitivity are understood as evaluable via backtracking, inductive knowledge is sensitive. (And moreover, there is nothing suspicious about backtracking counterfactuals.)
In response to the claim that sensitivity is incompatible with knowledge of merely highly likely events, I argue that we do not know those things. (And if quantum mechanics implies that the entire universe functions like an enormous lottery, then quantum mechanics implies that we know less than we ordinarily think.)
After defending and arguing for the standard sensitivity condition, I go on to argue for an even stronger form of this condition. By defending the viability of simple causal or counterfactual accounts of the epistemic basing relation, in combination with the claim that believing a proposition often amounts to believing many more fine-grained things, I argue that knowing that p requires that for every q such that one believes p partly by believing q, if q were false then one wouldn’t believe q (on the same basis as one actually does). This condition is able to explain a wide range of cases, and solves the well-known generality problem for process reliabilism.
Finally, I discuss how the sensitivity condition interacts with other popular positions in epistemology (safety, virtue epistemology, hinge epistemology, and contextualism). The sensitivity condition is compatible with all of these positions and makes valuable contributions to our understanding of epistemic luck and contextualism.