“If you torture the data long enough,” the saying goes, “it will confess to anything.” Although this is a problem for scientists, the stakes are higher for torturers. If tortured people really will tell you anything, how do you know when they are telling the truth?
Why Torture Doesn’t Work has a specific origin, says its author Shane O’Mara, professor of experimental brain research at Trinity College Dublin in Ireland. In 2009, he read an article about the release of the “Torture Memos”, legal documents prepared for the US federal authorities on the use of waterboarding, sleep deprivation, binding in stress positions, and other “enhanced interrogation” techniques.
Morality aside, O’Mara wanted to know if there was credible science that showed torture worked. The answer, it turns out, is no. The reality is that “the intelligence obtained through torture is so paltry, the signal-to-noise ratio so low, that proponents of torture are left with an indefensible case”. Advocates defend torture with an “ad hoc mixture of anecdote, cherry-picked stories and entirely counterfactual scenarios”, he says.
My review of Shane O'Mara's new book, Why Torture Doesn't Work, appeared in the November 11, 2015 issue of New Scientist.