Torture is a dangerous innovation; it would appear that it is an assay not of the truth but of a man’s endurance. . . . . What would you not say, what would you not do, to avoid such grievous pain. Pain compels even the innocent to lie.
Michel de Montaigne (1533-1592): On Conscience (trans. M.A. Screech – Penguin Classics)
I felt it worthwhile republishing a pair of blog posts from two years ago after reading, in last Wednesday’s Guardian, about a flawed piece of research suggesting that support for torture is widespread in the States. The book I refer to in the posts – Darius Rejali’s Torture and Democracy – conclusively demonstrates, at least to my mind, that no form of torture will ever be effective no matter how acceptable we manage to persuade ourselves it is.
The previous post has looked at recent Amnesty International survey findings concerning UK public opinion about torture. It drew on Darius Rejali’s Torture and Democracy specifically his chapter looking at whether torture works. It dealt with the problem of whether a torturer would know if you were lying. It seems that disturbingly often he would not. This post is examining the other criteria by which we might decide whether torture is effective and can contribute to the acquisition of life-saving information: the list picks up from the previous post with what is the second criterion in the overall list.
We had already established that, for me, there are insuperable moral objections to torture, but there are many people for whom it would be an option if it were effective. I hope they are reading this.
B. How well do cooperative prisoners remember?
Even the most elementary understanding of the effects of stress upon memory might cause a modern Torquemada or Topcliffe to question how useful it is to frighten someone out of their wits in order to obtain accurate and detailed factual information. Rejali is not slow to build on this perspective (page 466):
Lawrence Hinkle, a neurobiologist who, along with Wolff, advised the CIA on brainwashing, put it bluntly: “any circumstance that impairs the function of the brain potentially affects the ability to give information, as well as the ability to withhold it. . . . . The subjects’ ability to recall and communicate information accurately is as impaired as his will to resist.”
Sheila Cassidy is an interesting case in point again (page 467):
[She] could not remember information she knew only weeks before the arrest even when she wanted to cooperate. After days on the electric parilla, “I found it quite impossible to lie for the shocks came with such frequency and intensity that I could no longer think. So they broke me.” She decided to tell them the street address of where she had medically treated the man they wanted, but she could not remember.
This is probably an example of Ribot’s gradient which describes how, ‘when there is trauma to the brain, the farther back the memory, the more likely it is to survive the trauma.’
The conclusion is sobering for any aspiring torturer who cares to read about it (ibid.):
Ribot’s gradient does not occur for every kind of torture, but many techniques that cause pain also damage memory. By the time prisoners wish to co-operate, it may be too late.
Rejali then describes how a combination of sleep deprivation and repeated questioning under basic torture combine to create (page 468) the ‘illusion of knowing.’
Under repeated questioning, sleep deprived subjects display ‘higher confidence, but not greater accuracy.’
His final conclusion on this question is damning (page 469):
. . . . prisoners who experience memory lapses from trauma are likely to remember fictions quite vividly. When combined with the typical unskilled torturer, all this is a recipe for retrieving extremely poor information. Because both interrogators and cooperative prisoners have high confidence in this information, others are unlikely to second-guess its accuracy. This is yet another manner in which torture produces mistaken information.
C. How Good is the Information Extorted?
This moves us onto the next question.
The evidence marshalled to address the first two questions is already pointing us towards a sceptical position on this one as well. Rejali writes (page 469):
. . . . there is statistical evidence that interrogators produce consistently poor information in battlefield conditions or emergencies involving torture.
He then, as his first example among several, analyses in detail the efficacy of the Phoenix Programme, an initiative of the CIA during the Vietnam War designed to capture and kill Vietcong operatives. They had devised what they clearly considered a careful process of verification before anyone was finally caught and killed.
Its success rate was not impressive. Kalvyas and Koch, two researchers who had access to the CIA database for research purposes estimated (ibid.):
The Phoenix program ‘victimised at least 38 innocents for every 1 actual Vietcong agent.’ . . . . This is truly awesome perverse selection.
It is plain as a pikestaff that the information they were acting on was seriously flawed. Rejali feels that poor information entered the database in two ways (page 471): ‘in the selection of people and in the methods of interrogation.’ He explains (page 472):
The database suggests that the pool of individuals selected was composed of people with no information to give. Informants may have voluntarily fingered some, while others undoubtedly talked under torture. . . . . [I]t seems plausible that torture compounded the selection errors: the ignorant figured the innocent and deceived the torturers, and the innocent were then interrogated or terminated.
By contrast, he adduces a most telling example from World War II (page 476-77):
. . . . during World War II, British counterespionage managed to identify almost every German spy without using torture – not just the 100 who hid among the seven thousand to nine thousand refugees coming to England each year, not just the 124 who arrived from friendly countries, but also the 70 sleeper cells that were in place before 1940. Only three agents eluded detection; five others refused to confess. The British then offered each agent a choice: talk or be tried and shot. Many Germans chose to talk and became double agents. They radioed incorrect coordinates for German V missiles, directing them to land harmlessly in fields. But for this misdirection, the distinguished war historian John Keegan concludes, in October 1944 alone the Germans would have killed about 1,300 people and injured 10,000 others.
There is no reason to think that torture, or even the fear of torture, would have yielded any better results.
I am inclined to agree with his conclusion. I accept that I do not know how many others might also be concealed behind the word ‘almost:’ what is clear is that very significant numbers of lives were saved. I accept that the threat of death would not pass my ethics test but here we are merely concerned with the pragmatics, ie ‘Does torture work?’ Clearly it doesn’t under any of the circumstances delineated so far.
D. What about when time is short?
Rejali contends that an emergency with its condensed time frames does not alter the limitations described above. In fact (page 474) ‘it intensifies them by limiting what torturers can do.’ He explains that ‘real torture – not the stuff of television – takes days, if not weeks.’ Moreover, the techniques that could break someone by crossing their pain threshold quickly ‘[r]isk brain trauma. Ribot’s gradient becomes especially troublesome in ticking time bomb cases, since here the memory torturers want to extract is a recent one.’
Where does all this leave the pragmatic torturer?
I can’t do better at this point than quote at length from Rejali (page 478):
For harvesting information, torture is the clumsiest method available to organisations, even clumsier in some cases than flipping coins or shooting randomly into crowds. The sources of error are systematic and ineradicable. Innocent and ignorant prisoners generate malicious information, using torturers to settle private scores. Only highly experienced interrogators can spot such deception. Cooperative prisoners are unlikely to remember well and may give false answers with confidence. Neither they nor the interrogators can easily detect these errors.
In short, organised torture yields poor information, sweeps up many innocents, degrades organisational capabilities, and destroys interrogators. Limited time during battle or emergency intensifies all these problems.
In fact, the only time when torture might work well is ‘when we need it least, peacetime, nonemergency conditions.’ There is a catch though (pages 478-79):
Even then, torture has problems that cannot be eliminated, including desensitisation, death, unconsciousness, the loss of memory caused by damage, and the production of information that is more reliable the more it pertains to the remote past, not the immediate present.
Even if we number amongst those who feel that the ends can justify even the most revolting of means, it would be hard to continue justifying torture in the light of this evidence.
‘Homo sum, nihil humanum mihi alienum puto.’
[I am a human being; I believe that nothing human is alien to me.]
This cuts two ways.
The first part corresponds to the Golden Rule which in the Bahá’í Writings is expressed as, ‘Lay not on any soul a load which ye would not wish to be laid upon you, and desire not for anyone the things ye would not desire for yourselves.’ In effect this means, ‘I must see that you are the same as I am and treat you as I would wish to be treated myself.’
However, the quotation from Terence also forces us to consider another possibility: ‘Even though I hold what you do to be evil, I am as human and flawed as you are and might do the same myself in your situation.’
The findings of Amnesty International are disturbing for two sets of reasons. First of all, they show that many in our society can contemplate treating others in ways they would never wish to be treated themselves. Secondly, they force me to recognise that the society with which I most closely identified as a child and a young adult is seriously flawed in terms of universal support for its supposedly most cherished values. And that is a dangerous place for any liberal democracy to be in.
We do indeed have a long road to travel as a society before we move from being industrially advanced to being truly developed.