Innovation
Technology
The technology utilizes a standard, non-intrusive video-based eye-tracker which tracks the point of gaze on the screen. In addition, several specially-developed machine-learning algorithms enable accurate classification of familiar and unfamiliar faces.
When a subject is shown photos of possible acquaintances, the gaze behavior is analyzed, revealing the following patterns:
-
Familiar faces initially attract the viewer’s fixation before moving to a different face. This fixation is followed by a strong tendency not to return to it (the ‘repulsion effect’). During the first second of observation gaze is directed significantly more towards familiar faces, than towards unfamiliar faces.
-
Unfamiliar faces During the whole duration of observation, gaze is directed significantly more towards unfamiliar faces, than towards familiar faces.
The technique has an accuracy rate of over 91% for detecting familiar faces.
Furthermore, the technique demonstrates a higher detection efficiency in differentiating between participants that are familiar with one of the faces and participants that are unfamiliar with all of the faces. This classification accuracy exceeds all traditional CIT based on autonomic (e.g. skin conductance) as well as neuroimaging measures.
Benefits
-
More accurate technique for detecting deception, compared to measuring other physiological responses such as autonomic and neuroimaging measures
-
Portable, convenient method that is quicker than monitoring other physiological parameters, such as increased skin conductance and neuroimaging methods
-
Could be applied without the tracked individual’s awareness
-
No need to connect any peripheral equipment, such as electrodes
-
Can be implemented by anyone, requiring minimal technical training
Development milestones
-
The technology was implemented in two experiments, in which photographs of human faces were viewed. In these experiments, participants were requested to memorize several faces, including personal Facebook friends of theirs, and were instructed to try and conceal familiarity with any faces they recognized in the photos.
-
Future research will include neuronal and cognitive mechanisms underlying the familiarity-related ‘attraction’ vs. ‘repulsion’ effects, and investigating the correlation between LTM (long-term memory) and attention.
-
Additional research will focus on tracking eye movements using other stimuli, such as photos of geographic locations or objects, combined with other measurable physiological responses.
-
We will also examine whether this technique – as most physiological measures – is vulnerable to countermeasures.
-
So far GIF young grant has been received.
Applications
-
Secret services, intelligence and security agencies
-
Airports
-
Police agencies and Interpol
PATENT STATUS
Granted US 11020034
