This page is an evolving compilation of my research on lie detection, intended primarily as a resource to other researchers. Most recently I’ve done some work on Converus in the US and iBorderCtrl in the EU. My doctoral dissertation and post-doctoral research on bias in decision-making technologies (esp. polygraphs) and policing was funded by the National Science Foundation. I have researched policing, technology, and their intersections in lie detection for over 10 years, collaborating with a variety of investigative journalists, researchers, activists, and others. Applications include international security and border security.
Recently I gave a related talk on “Psychic X-rays: Holy Grail, Cover Story – or the End of Human Rights?” (video; slides) as part of the Border (Dis)placements symposium that took place 13 May at Stroom Den Haag. I argue we should keep current bans on algorithmic decision-making and profiling, ban mass security screenings for low-prevalence problems, and recognize cognitive liberty as part of human dignity.
The National Academy of Sciences (2003) lit review endures. It’s not going to be outdated anytime soon, because the scientific consensus is that lie detection doesn’t exist. So there is no vibrant culture of peer-reviewed lie detection science.
Graduate Work (2010-2014)
My 2014 University of Virginia Ph.D. dissertation, supported by an NSF Doctoral Dissertation Research Improvement Grant among other awards/honors, showed that bias can seep into technology-mediated processes; but we also have to worry about the qualitative ways that values set the terms of these processes themselves. I subsequently came to admire Alex Vitale’s work on policing and Virginia Eubanks’ on surveillance of the poor, and would certainly situate mine in that context if I ever went back to it.
Here’s an off-shoot from the medical decision diagnosis comparative case study: “Overriding Race and Class Bias: Equity in Technology-Mediated Diagnosis.” It shows that even much more scientifically moored, obviously useful tools such as medical decision-making tools can be vulnerable to bias.
Here are the results of a survey of Virginia state-licensed polygraphers that I conducted as part of my pre-doctoral research for two graduate methods classes: “Virginia Polygrapher Survey 2010-2011.” They indicate that polygraphers are demographically like law enforcement management: overwhelmingly white, male, older, conservative, and current or former law enforcement—and that one in five of them reports believing that some groups of people (e.g., blacks) are more likely to pass or fail the polygraph than other groups.
Ft. criticism in Ann-Kathrin Nezik’s Die Zeit article “Falsch geblinzelt,” Aug. 2020.
Ft. criticism in Jake Bittle’s MIT Technology Review article “Lie detectors have always been suspect. AI has made the problem worse,” Mar. 2020.
“Protecting Sacred Space: Real ‘Lie Detection’ Would Threaten Human Rights As We Know Them,” a moral first principles argument against (1) lie detection, and (2) critiquing attempted/purported lie detection on the basis primarily of its insufficient scientific basis (as I and others have sometimes done), iBorderCtrl.NO, June 2019.
“Good Logic, Bad Life: Decision-Making Neuroscience Suggests AI Decision-Making’s Weaknesses,” an application of Damasio to AI decision-making contexts more broadly, iBorderCtrl.NO, May 2019.
“Why Most Published iBorderCtrl Research Findings Are Likely to Be False,” an application of Ioannides to iBorderCtrl, iBorderCtrl.NO, May 2019.
“iBorderCtrl Automates Discrimination,” a critique of avowed neutrality in iBorderCtrl, iBorderCtrl.NO, Jan. 2019.
“Biomarkers of Scientific Deceit,” a critique of automated deception detection through micro-expressions in iBorderCtrl, iBorderCtrl.NO, Jan. 2019.
Ft. criticism of next-generation “lie detection” tool Converus in Mark Harris’ Wired article “An Eye-Scanning Lie Detector Is Forging a Dystopian Future,” Dec. 2018.
“Five Posts on Polygraphs: A Summary,” Oct. 2018.
“Warning Shots of Corruption: Releasing and Revisiting the Polygraph Interviews that Launched My Dissertation Research.” AltGov2, Oct. 2018.
Ft. research in Mark Harris’s Wired article “The Lie Generator: Inside The Black Mirror World of Polygraph Job Screenings,” Oct. 2018.
The Science Creative Quarterly extension of problems with polygraph programs to the broader contexts of mass surveillance and refugee screening. This is a simple extension of the Bayes’ Rule application from the NAS 2003 polygraph report that shows mass screenings for low-prevalence problems have the potential to harm security and human rights alike. Here is a precursor reproducing the NAS application and explaining its implications.
Curated re-release of federal polygraph documents that I made open-source, that then disappeared… Suggesting that I should perhaps put my research on my own website this time. (Hi.) I shared related files, including many obtained through years of Freedom of Information Act (FOIA) requests and FOIA lawsuits against several federal agencies, interviews, and other muckraking with McClatchy for a multi-part national investigative series here. The lawyer and reporter in this case did not use encryption as a general practice or upon my request, and the documents I made open-source later disappeared from both their websites. So one lesson there is to use good infosec and work with people who do.
Another note on FOIA: The federal system is broken. Even with NSF support for my dissertation research, I never got the data I needed to test for bias in federal polygraph programs. Instead I paid a lot of money to lose. Except I eventually “won” a Pyrrhic Freedom of Information victory for American educational requestors, so you can play the same game too? Don’t. It’s a losing game. Especially if you have a defined research agenda with specific hypotheses you’d like to test with specific data—unlike a journalist who can just go fishing for interesting stuff and hope to get lucky—FOIAing state and local agencies in the U.S. is generally a better bet.
A paper I was honored to present by invitation at an American Studies Symposium at Osaka University in Japan (2016), alongside keynote speaker/lie detection history rockstar Ken Alder: “Lie Detectors On Trial? Science, Security, and Accountability in the Era of the Hague Invasion Act.” It reformulates polygraph programs as an interesting corruption problem more than anything else, since the scientific consensus has basically always been against them—yet the programs grow and grow.
As I wrote for AltGov2, my dissertation research began as a series of interviews about polygraphs that I filmed, intending to make a documentary on the subject. But then I wanted to know the scientific answers to the questions that the anecdata suggested, so I wrote a dissertation trying to find out. Now some of the people I interviewed at the start of this process are dead (of natural causes), and they generously gave me their time while they were alive… So it seems right to share with the world what they shared with me. These interviews are licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License. Thanks again to all my interviewees.
A note on transcripts: I tried recompiling transcripts using Youtube’s closed captioning feature (click video in video manager, click subtitles/CC, set language to english, hit english, actions —> download SRT) with this captions converter (srt to txt) for clean-up. But I was unhappy with the quality of these transcripts, and cleaning them up is just too much work. For the moment, I have opted to refrain from posting what I have in the way of transcripts since if it were me as a subject, I would want it to be right.
A note on tapes: Some of the tapes themselves must be re-rendered from original NTSC DV due to damaged or missing files, and I’ve had difficulty finding a place to digitize that American format in Berlin. Contact me if you have old-school HD equipment in Berlin and you are a nice person. Please forgive me if your interview is currently missing from this archive as a result of this technical snag.
Stephen Fienberg. Video.
Witness to Innocence—death row exonerees Shabaka Waqlimi (aka Joe Green Brown), Clarence Brandley, Herman Lindsey, Derrick Jamison, Juan Melendez, and Dave Keaton. In memoriam Dave Keaton and Mamie Brown. Video (missing a tape/in progress)
Tim Longo. Video.
Rose McDermott. Video.
Documents / Open records request releases
Due to relevant documents’ disappearance from the national newspaper and lawyer websites where I had made them available—a disappearance explained as well as possible and partially remedied through attempted re-released here—it does not appear to be possible to show where and how these documents were originally published.
Nonetheless, it looks like here is a partial archived grouping of documents originally I obtained through open records requests and subsequently provided to McClatchy Newspapers in 2012.
Here is a 2016 NSA release responding to a 2011 open records request regarding polygraphs and equal opportunity (added to this compilation Nov. 2018).
Here are the briefs and oral arguments from my polygraph open records request case that Judge Kavanaugh decided (added Nov. 2018).