Sunday, October 29, 2006

Bishop and Trout: Epistemology and The Psychology of Human Judgment

We've started reading the above book in my epistemology course. I'm quite a fan of it, liking the authors' discussion of "Standard Analytic Epistemology" and their replacement project which can be called either "ameliorative psychology" or "applied epistemology." It's a really good book, and a complete breath of fresh air.



Bishop and Trout make a big deal of "Statistical Prediction Rules" or SPRs that, they argue, outperform experts on particular tasks. So, for instance, there's an SPR that tells you whether a convict is likely to commit another crime. And these SPRs are usually better at these kinds of questions than the experts.

Of course, no one likes to hear this. We like to think we're better than a stupid rule, and we'd like to think that we have special insights that make us more reliable than, say, a neophyte who follows a rule, but doesn't really understand. Sadly, we may be wrong about this. As Bishop and Trout argue, when people deviate from an SPR (thinking they have additional knowledge that undermines its validity in a particular case) they usually do worse than if they'd continued following the SPR.

But several good ideas did come out of our discussion. One was this: while we may not object to SPRs on epistemic grounds, we may object to them on other grounds. Consider this: perhaps there's an SPR that will tell you who's most likely to hijack an airplane. And that SPR may say that you should use racial profiling: don't waste your time on the grandmothers; instead focus on the young unmarried men of Arab descent. Would we be justified in using the SPR? The thought in class was that there would be other considerations that would trump the epistemic value of using the SPR: specifically, considerations of justice and fairness. That seems like a fair point to me.

Thus, while we may be wrong to defect from an SPR on epistemic grounds, we may be justified defecting on other grounds, including grounds of justice and ethics.

0 Comments:

Post a Comment

<< Home