FAS

Quest for Truth: JASON on Detecting Deception (2008)

03.01.21 | 3 min read | Text by Steven Aftergood

Is it possible to reliably detect persons with malicious intent by remotely measuring their physiological state or studying their involuntary verbal or behavioral signs?

That is a problem that is far from having a technological solution, the JASON science advisory panel told the Department of Defense in a 2008 report that was released under the Freedom of Information Act last week. See Quest for Truth: Deception and Intent Detection, JSR-08-143, October 2008.

Assessing intent is “an enormous challenge for the DOD due to the complexity of human behavior, as well as the cultural variances in the ways a person can express his or her intent and the inducements leading to implementation of that intent.”

“When it comes to the detection and assessment of covert combatants’ intent, the consequences of cultural differences may be insurmountable without years of mentoring, training, and education,” the report said. Detecting intent “is simply not feasible today through technology, interview techniques, or coupled interview and technology solutions, even within the same culture.”

The report reviewed a range of biometric technologies that could remotely and even covertly measure physiological changes for assessment. Most of them were found to be unproven for purposes of deception detection, and none of them were mature. These included Laser Doppler Vibrometry (to measure heart rate and heart rate variation), radar, thermal imaging (to monitor changes in skin temperature), and eye tracking.  In incident-specific queries, the polygraph can distinguish lies from the truth “at rates well above chance,” but it is not useful for screening or remote application.

Voice stress analysis (VSA) is one particularly “useless” approach, the report said. In a sarcastic aside, the JASONs said that “Installation of a cardboard box covered in aluminum foil would be far cheaper that the purchase of a VSA system, but still could be used to make subjects believe that a functional lie detector is in place, thereby improving the likelihood that subjects will tell the truth.”

At least as of the time of the report, “No compelling evidence yet exists to support remote monitoring of physiological signals in an operational scenario.” And “No scientific evidence exists to support the detection or inference of future behavior, including intent.”

The JASON authors also said that any further pursuit of research in this area needed to be firmly grounded in empirical research and validation.

Disclosure of the unclassified JASON report had been categorically denied under the Freedom of Information Act in 2010, but it was released on February 25 with only minor redactions following a decade-long appeal process. (The name of one researcher was withheld on privacy grounds).

*    *    *

Last month the Department of Defense fully denied another FOIA request filed in 2015 for multiple unclassified JASON reports dating from 2013 and 2014.

All “528 pages are withheld in their entirety [as] ‘technical data with military or space application in the possession of, or under the control of, the Department of Defense’,” according to the February 16 denial letter. The denial was appealed.

*    *    *

Last week the Census Bureau released a JASON report from last month on how to improve census data quality.

The census doesn’t have to be (and actually can’t be) perfect, the JASONs said, but it can be improved.

“It is important to understand — and for the Census Bureau to communicate to the public — that the accepted range over previous decades allows for considerable imperfections, as long as these do not knowingly embody a priori biases against individual states or statutorily defined classes of individuals.”

See Assessment of 2020 Census Data Quality Processes, JSR-20-2N, February 8, 2021.

publications
See all publications
Emerging Technology
day one project
Policy Memo
Strengthening Information Integrity with Provenance for AI-Generated Text Using ‘Fuzzy Provenance’ Solutions

By creating a reliable, user-friendly framework for surfacing provenance, NIST would empower readers to better discern the trustworthiness of the text they encounter, thereby helping to counteract the risks posed by deceptive AI-generated content.

02.13.25 | 7 min read
read more
Education & Workforce
day one project
Policy Memo
Tending Tomorrow’s Soil: Investing in Learning Ecosystems

By investing in the mechanisms that connect learning ecosystems, policymakers can build “neighborhoods” of learning that prepare students for citizenship, work, and life.

02.12.25 | 6 min read
read more
Global Risk
Blog
Don’t Let American Allies Go Nuclear

Empowering U.S. allies to do more so Washington can do and spend less sounds attractive. But enabling, or looking the other way at the spread of nuclear weapons is not in America’s interests anymore today than it was in the 20th century.  

02.12.25 | 5 min read
read more
Government Capacity
day one project
Policy Memo
Protecting Infant Nutrition Security:
Shifting the Paradigm on Breastfeeding to Build a Healthier Future for all Americans

Policymakers on both sides of the aisle agree that no baby should ever go hungry, as evidenced by the bipartisan passage of recent breastfeeding legislation and widely supported regulations. However, significant barriers remain.

02.11.25 | 10 min read
read more