### Monday, August 27, 2007

## Knowability and Actuality

Here's a point which dropped out of my last post. Anti-realists adopt some local or global version of the knowability principle:

KP: P only if it is possible that P is known.

They've understood this principle such that it's constrained by the actual distribution of truth-values; the actual distribution of truth-values over statements determines what is knowable. Why?

In my post, I took it to be because otherwise knowability isn't nearly demanding enough. Suppose that P is false. Still, P might be knowable in the sense that there is some world in which P is knowable - it might only have to be a world in which P is true. But the anti-realist was looking for a more substantial sense of knowability, one according to which falsehoods are not knowable. So anti-realists understand knowability in such a manner so as P is knowable if and only if P.

In the Analysis note I discussed in the previous post, Jesper Kallestrup clearly sees things quite differently. He writes:

'It would be too easy to refute anti-realism if [KP] failed in cases where P is actually true, but not known in worlds in which P is false.'

The suggestion is that having what's knowable determined by the actual distribution of truth-values blocks putative counterexamples to KP. But I'm not seeing how the point is supposed to go. Suppose that P is actually true, but not known in worlds in which P is false. I'm not seeing any danger from such cases, though they refute the following thesis:

KP!!: P only if it is necessary that P is known.

Since KP doesn't require that P is known in all possible worlds, it's not clear to me how the observation that P is not knowable in worlds in which P is false is meant to provide the material for an even prima facie counterexample to KP. So it's quite unclear to me how consideration of such cases could motivate adopting the thesis that what is knowable is fixed by the actual distribution of truth-values. Unless I'm missing something, I think the motivation has to stem from the worry about diluting the notion of knowability too much.

(Andreas blogged about related issues a while back).

KP: P only if it is possible that P is known.

They've understood this principle such that it's constrained by the actual distribution of truth-values; the actual distribution of truth-values over statements determines what is knowable. Why?

In my post, I took it to be because otherwise knowability isn't nearly demanding enough. Suppose that P is false. Still, P might be knowable in the sense that there is some world in which P is knowable - it might only have to be a world in which P is true. But the anti-realist was looking for a more substantial sense of knowability, one according to which falsehoods are not knowable. So anti-realists understand knowability in such a manner so as P is knowable if and only if P.

In the Analysis note I discussed in the previous post, Jesper Kallestrup clearly sees things quite differently. He writes:

'It would be too easy to refute anti-realism if [KP] failed in cases where P is actually true, but not known in worlds in which P is false.'

The suggestion is that having what's knowable determined by the actual distribution of truth-values blocks putative counterexamples to KP. But I'm not seeing how the point is supposed to go. Suppose that P is actually true, but not known in worlds in which P is false. I'm not seeing any danger from such cases, though they refute the following thesis:

KP!!: P only if it is necessary that P is known.

Since KP doesn't require that P is known in all possible worlds, it's not clear to me how the observation that P is not knowable in worlds in which P is false is meant to provide the material for an even prima facie counterexample to KP. So it's quite unclear to me how consideration of such cases could motivate adopting the thesis that what is knowable is fixed by the actual distribution of truth-values. Unless I'm missing something, I think the motivation has to stem from the worry about diluting the notion of knowability too much.

(Andreas blogged about related issues a while back).

Labels: Anti-Realism, Knowability