IF THE NOSE liars are put forth as that of Pinocchio, policemen and magistrates would be happier, lawyers perhaps a little ‘less and legal thrillers would end after the first few pages. But unfortunately it is not. Today the analysis of the lie takes roads impassable and the results often questionable: the good old interrogation tactics dense verbal formalized once and for all by John Reid, or the analysis of the face encoded in Facial Action Coding System by psychologist Paul Ekman ( inspirational series Lie to me), or the lie detectors that they discover who bluffs by tiny changes in heart rate, pressure, breathing bioelectricity skin. O sensors that measure the reaction times of the brain in the face of questions, such as the implicit association test of neuroscientist Giuseppe Sartori, also used in the process Franzoni. All of these systems require the presence of an expert in the flesh. Today’s news, however, is a software that does it all. They developed computer scientists at the University of Michigan under a project funded by DARPA, the scientific research department of the Pentagon. It is a system that observes the video of testimonies and interviews and – through pattern recognition and machine learning – can find out who is lying in 75% of cases. Experts human instead settle only on a more modest 53%, according to a meta-analysis of 253 studies developed by psychologists at the University of California. “The difference is explained by the difficulties of the investigators in the human track of all the minor clues: for example, no one can count all the times that the interviewee uses the pronoun” I “or looks upwards, while the software does “says Rada Mihalcea, leader of the project. The meticulousness Hound digital is not innate, but learned and then continually improved: Mihalcea and colleagues have studied the system shooting 120 processes, the performance of defendants and witnesses have been broken down into the sum of all their gestures and the software associated with the frequency of each gesture knowledge, in retrospect, the outcome of the process. So to understand how they behave statistically liars. The next step? The establishment of a database continuously updated, a sort of “gestural alphabet of lies”, searchable by law enforcement. What are the signs of insincerity that are arching an eyebrow suspiciously virtual American software? Grimaces and frowns, for example, are used by 30% of the 10% of liars and honest. A look directed toward the interrogator may be slightly more indicative of lies (70%) and sincerity (60%). Gesturing with both hands is more typical of lies (40%) than for the truth (25%). Use the third person singular more than the first, as to distance themselves from what is being told, it is more liars from honest. Altrettando false, according to computer analysis, it is who uses too many interjections as “uhmmm”, “ehhhh” to fill the spaces between sentences. It will not help that much more, however, learn by heart all these telltale signs to train yourself to not lasciarseli escape. Experts from the University of Oulu (Finland) have in fact just made the first algorithm of computer vision that recognizes the micro expressions, that those signs – imperceptible to the naked eye and can not be controlled consciously – that appear on our face when we try to hide something in situations voltage.
- Arguments:
- software
- Lies
- Starring:
No comments:
Post a Comment