Laurence and Emily Alison had access to something almost nobody in this field gets: recordings of real interrogations, of real terrorism suspects, in real cases, conducted by police in the United Kingdom. Not students in a laboratory, not mock crimes — 181 actual interviews, and in later work over a thousand, with suspects from Al-Qaeda-inspired, Northern Irish and extreme right-wing cases.
They coded the interviews twice. Once for the interviewer's behavior, using a framework drawn from Motivational Interviewing — acceptance, empathy, autonomy, evocation — and from interpersonal circumplex theory, which describes behavior on axes of authority and affiliation. And once for the suspect's output, using a scale of counter-interrogation tactics ranging from complete silence through monosyllabic responses and scripted refusals to full engagement, and a measure of how much investigatively useful information was actually obtained.
The results are consistent and directional. Adaptive interviewer behaviors — empathy, acceptance, respect for the suspect's autonomy, evoking rather than lecturing — predicted increased disclosure. Maladaptive behaviors — confrontation, judgment, dominance, cutting off — predicted increased use of counter-interrogation tactics and reduced yield.
The mechanism the Alisons identify is not that suspects come to like the interviewer, which would be a naive reading. It is reactance.
Brehm's theory holds that when a person perceives a threat to a behavioral freedom, they experience a motivational state directed at restoring it, and the threatened behavior becomes more attractive. A terrorism suspect who has been trained to resist interrogation, and who understands their silence as a principled act, is maximally reactant. Pressure supplies exactly what their resistance needs: an adversary, a confirmation that the interview is a contest, and a freedom to defend.
Autonomy-preserving questioning removes the trigger. If the interviewer explicitly acknowledges that the suspect may choose not to answer, there is nothing to resist. The Alisons' data show that this is not a rhetorical nicety — respect for autonomy was among the strongest predictors of disclosure in their coding.
There is a second finding that matters as much and is easier to act on. The relationship between rapport and yield was not linear in the way one might assume: a single instance of maladaptive behavior could undo a great deal of accumulated rapport. Confrontation was more damaging than empathy was helpful, which means the discipline required is negative — not doing the thing — rather than performative.
Two things make this study unusually strong evidence. It is field data with real stakes and real consequences, not a simulation. And it was conducted on the population most resistant to persuasion that exists, which is where a rapport-based method should fail if it were merely a pleasant approach that works on cooperative people.
It did not fail. It outperformed, on exactly the suspects who were trained to defeat interrogation.
Cross-ref: Senate torture report chapter