The section that most caught my eye was the discussion of polygraph countermeasures, and particularly a section on a fellow, who after being wrongly convicted for murder on polygraph evidence, took it on himself to hack the polygraph test to help prove his innocence, all while being wrongly imprisoned.
The most famous countermeasures test was probably conducted by Floyd ‘Buzz’ Fay, a man who was falsely convicted of murder in the USA on the basis of a failed polygraph examination. He took it on himself to become a polygraph expert during his two-and-half years of wrongful imprisonment. He coached 27 inmates, who all freely confessed to him that they were guilty, in how to beat the control question polygraph test. After only 20 minutes of instruction, 23 of the 27 inmates were successful in defeating the polygraph examination.
The report discusses empirical evidence on how well these tests detect potential mistruths (not brilliantly it seems) and contains summaries of research which shows the percentages of hits and misses each sort of test is likely to make.
For example, in a form of polygraph test known as the Control Question Test (where responses to direct questions about the crime are compared to responses to indirect questions) over 26% of innocent suspects were scored as lying, although in the Guilty Knowledge Test (where responses to items of information only a guilty person would know are compared to responses to other information) only 4% of innocent suspects were wrongly scored as lying, but guilty suspects were correctly identified only 59% of the time.
Link to BPS report on ‘Polygraphic Deception Detection’.