Recently we blogged about the advertising of the future now new research has upped the ante with computers that can catch you telling lies.
Has it really happened already?
Past research has found that the average person can only delineate a lie from the truth about 50% of the time. It has also shown that experienced experts can do so only about 65% of the time. Then there are a rare few dubbed “Truth Wizards” that are capable of reading lies with a 80% accuracy rate.
The Scientifc American has reported that researchers at the University of Buffalo have created video software that they claim can aid interrogators/interviewers in detecting deception.
Their software analyzes eye movement for signs of deceit and according to the researchers can accurately determine deception 82.5% of the time.
Dr. Mark Frank a seasoned lie detector, University of Buffalo professor of communications, and co-author of the study believes that their results have laid the foundation for larger studies to incorporate a greater testing group size and add the dimension of body language in an effort to determine whether new technologies can aide investigators with a significant percent of accuracy.
It is important to point out that this study did not focus on the myths of lie detection such as looking to the left or down constitutes a lie. They did have computer software track participants eye movements while being questioned. In the beginning of the interviews the interrogators focused on conversational questions to determine a baseline for each participant. Therefore, a deviation from this baseline could suggest there was more to what was being said.
The software compared each subject’s baseline eye movement with those observed during the questions about a check that the participants may or may not have taken. If the computer detected a large deviation, the researchers noted this change and flagged that person as a potential liar.
Joe Navarro a retired FBI counter-intelligence special agent commented on the need to take the whole body as well as the face into consideration when trying to detect deception, “I can tell you as an investigator and somebody who’s studied this not just superficially but in depth, you have to observe the whole body…”
The Buffalo researchers do plan to take a more holistic view of behavioral cues in future studies.
“We know that the eyes give signals that lead to deception, but what about general body movements,” states Ifeoma Nwogu study co-author and a research assistant.
Nwogu goes on to state that faster algorithms would also enable the software to flag behavioral deviations in near real-time.
What are your thoughts on this type of computer deception software?
A threshold question is the reliability of eye movement to indicate deception. Direction of gaze, standing alone, or blinking probably would not be an adequate basis to support a conclusion of deceit. The baseline idea is a good one, but I don’t feel that the automated systems are equivalent to human analysts just yet–the machines are too focused on one channel, whether this effort with the focus on eyes, or the voice stress detectors with a singular focus on voice.
80 percent is a really good and about the same accuracy as micro expressions with less training. The thing it that machines and people can be fooled but only machines can be fooled repeatedly and consistently by using the same method.
Peter- Interesting comment, it reminds me of Albert Einstein’s definition of insanity, ‘Doing the same thing over and over again and expecting different results.’
The thing to take into consideration in regards to computer programs is that they are constantly being upgraded to add depth in an effort to prevent that from happening.