A long time ago in a galaxy far far away (that’s the early 90s) Keith DeRose’s famous bank cases forever changed the thought experiment multiverse. Since their original inception, there’s been lots of well-known and exciting work developing these probes in some of our best competing theories of knowledge. And orthogonal this theoretical research, experimental epistemologists of the last few years have sought to investigate the factors that influence our ordinary language practices involving knowledge sentences [here] and [here] and [here] in part by also borrowing from these cases.
In a recent paper discussed [here on this blog] and [here on CD] Keith offers some really nice objections to earlier experimental work on bank cases regarding their bearing to epistemic contextualism. We got a chance to talk to him after his presentation at a recent [MERG Lab Meeting] at NYU about the best way to address his worries, and Josh Knobe and I have now developed a new study that brings together the two groups above by avoiding many of the problems in earlier research.
In this new study, we asked participants to make truth judgments about a knowledge assertion or denial made by a speaker in the actual vignette (as opposed to simply being asked whether a given character in that vignette has or doesn't have knowledge). We also used cases that manipulated both high stakes and a salience of error possibilities together (as opposed to just one or the other). This resulted in a 2 x 2 x 2 study that varied stakes (high vs. low), error possibilities (made salient vs. not) and the character's actual claim (knowledge assertion vs. knowledge denial).
A short write-up of these new experimental results is available here [Download Bank Results], but a quick snapshot is shown in the chart below:
The key message seems to be that pace previous experimental work, the factor of error possibilities may make some sort of impact on ordinary language practices after all. Specifically, this study shows that error can influence the way people evaluate the truth conditions of sentences either attributing or denying knowledge. However, similar to [earlier experimental findings] once again no such effect was found here regarding practical subject stakes.
I would love to hear any thoughts you might have about this further experimental work on the bank cases, as well as suggestions for the ways we can continue to better investigate ordinary knowledge practices. So be sure to join in the discussion over at Certain Doubts: http://el-prod.baylor.edu/certain_doubts/?p=1999!