Cancer researchers train dogs to sniff out cancer signs in urine. Dogs agree with the cancer tests 41% of the time. But, according to today's Strib, on one occasion all the dogs agreed that somebody's urine was bad, though the tests were negative. Scientists re-did the tests, and found cancer. So, for the most part, you use the lab tests to evaluate the dogs' reliability: if the dogs agree with the lab, the dogs are detecting cancer. But, once the dogs have been shown to have some promise, you can also use the dogs to second guess the lab tests -- and you can use the consensus of dogs as a different kind of marker than the judgment of a single dog. So there is no one platform of utter certainty on which to stand, and yet we can get better at diagnosing cancer. Welcome to the grown up world of "epistemology." (By the way, this doesn't work just with dogs.)
Posted by shea0017 at September 24, 2004 12:05 PM