Of Cripples

I have been present at the examination of women accused of witchcraft, and I can report that the evidence for their guilt consisted almost entirely of the testimony of neighbours — neighbours who had quarrelled with the accused, neighbours who had suffered misfortune and needed it explained, neighbours whose imagination had been furnished by the popular literature of demonic activity with all the details necessary to complete a convincing story. The women themselves sometimes confessed, under enough pressure, to things I found it difficult to believe any human being had literally done. The judges found these confessions satisfying. I found them depressing.

We are too easily convinced by what we expect to find. The man who has decided that the lameness of his cattle is witchcraft will interpret every circumstance as confirmation, and will discard, without quite noticing that he is discarding it, every circumstance that points another way. This is not a failure peculiar to the credulous — it is the default operation of the human mind, which seeks patterns, finds them everywhere, and cannot reliably distinguish between patterns that are in the world and patterns that are in the mind’s need for order.

“After all, it is rating one’s conjectures at a very high price to cause a man to be roasted alive for them.”

The question of whether witchcraft exists I leave to those with more certainty than I possess. What I am confident about is that the evidence typically adduced for it — the testimony of witnesses, the confessions of the accused, the correlation of harm and suspected malice — is precisely the kind of evidence that a sober epistemology would treat with the most caution, not the least. Human testimony is compromised by interest, by fear, by the desire to tell a satisfying story, by the difficulty of accurately remembering what one has seen, and by the native tendency to see what one has been taught to expect.

I have learned from this to hold my own certainties lightly, especially the certainties that are most widely shared around me. A belief that everyone holds is not therefore a belief well examined — it is a belief that has simply not yet encountered strong enough resistance. The tester of beliefs is difficulty, not consensus, and the beliefs most in need of testing are precisely those that consensus protects from testing.