Face Recognition Technology

This post is not exactly about a weird condition, but it is weird. One thing most people can do better than a computer is recognize a face. The technology is out there. According to an archived article in the UK Guardian, "some of Britain's 2.5 million CCTV cameras are being hooked up to facial recognition systems designed to identify known criminal." CCTV stands for closed-circuit television, which simply means that it does not broadcast publicly, and instead sends feeds to specific locations. The technology is mostly used for security -- an airport in Germany, Nevada casinos, and London's Borough of Newham's police CCTV, which attempts to identify convicted criminals from a database of mugshots. According to the Guardian article, Newham's automatic face recognition technology has failed to work even once, but it is supposed to work by measuring the peaks and valleys of the structure of a face and 

  "picks out a series of 'nodal points' and measures the distance between them. The distances between points are converted to a string of numbers which are then compared against the set of suspects' face numbers in the database. If there is an 80% chance of the match being right, the computer alerts an operator in the CCTV control room and displays the two facial images on screen for a human check. If the operator confirms that a suspect has been spotted, he contacts the police, who decide whether or not to take action. As few as 14 nodal points are enough for a good match."

According to a French technology website, face recognition uses mainly the following techniques:

 -- Facial geometry: uses geometrical characteristics of the face. May use several cameras to get better accuracy (2D, 3D...)

 -- Skin pattern recognition (Visual Skin Print)

 -- Facial thermogram: uses an infrared camera to map the face temperatures

 -- Smile: recognition of the wrinkle changes when smiling

Prosopagnosia

  That man who mistook his wife for a hat was real. As far as I can tell, so is prosopagnosia, a neurological disorder that impairs the brain's ability to recognize faces. The word roughly means "face non-knowledge" in Greek, and those afflicted are either born with the disorder or have a particular area of the brain damaged . The National Institute of Neurological Disorders and Stroke has this to say about prosopagnosia (ten-times-fast!):

 

"some will only have difficulty recognizing a familiar face; others will be unable to discriminate between unknown faces, while others may not even be able to distinguish a face as being different from an object. Some people with the disorder are unable to recognize their own face."

  Cecilia Burman says that that she is "'face-blind' in the sense" that she "cannot recognize people by their faces." In a somewhat confusing attempt to describe "what it can be like to live with" prosopagnosia , she compares faces to the surface of mostly flat rocks. Attempting to identify individuals based on facial structure is like trying to recognize and identify different rocks. Kind of difficult, right?

Eva
FredStenJohn

 Right? Prosopagnosians sometimes use other clues to identify individuals: voice, clothes, hair color and probably anything that isn't related to the face, I would think.  A common symptom is a difficulty in recognizing characters on television.  

The related Capgras delusion, on the other hand, is conscious facial recognition but with no accompanying emotional response, sometimes leading to the delusional belief that a relative or spouse has been replaced by an imposter (or a hat!). This leads me to my next post about panopticon-esque computer technology and the ability to recognize faces -- something which computers have trouble with as well.

False Memories

Apparently, even your memory is out to deceive you. We hear all the time that our senses can deceive us, but I have never really understood the impact of that statement. Sure, I might mistake a bird for a plane, or maybe confuse a small rodent foraging for food outside my tent with bigfoot, but when it comes to actual events, that's a whole different ballgame. I mean, if I remember an event, like say hugging Bugs Bunny at Disneyland, then it must have happened. I might erroneously recall his fur as being purple, but that's about the extent of my brain playing tricks on me. 

Aha! Not so fast! Bugs Bunny is a Warner Brothers character, and if he was running around Disneyland, it was probably some sort of guerilla marketing scheme. Anyway, UC Irvine psychologist Elizabeth Loftus managed to implant that memory in some of her subject's memorybanks in a study.  Here is an excerpt from a CNN.com article detailing the study:

 "Loftus talked with subjects about their childhoods and asked not only whether they saw someone dressed up as the character, but also whether they hugged his furry body and stroked his velvety ears. In subsequent interviews, 36 percent of the subjects recalled the cartoon rabbit."

Her work is somewhat controversial. Loftus co-wrote "The Myth of Repressed Memory," a book whose title alone conflicts with the well-established Freudian conception of deeply-repressed memories of traumatic experiences that must be painstakingly surfaced through psychoanalysis. 

Loftus hints that an easy way to suggest the reality of an event that did not occur is too use sensory detail -- sights, sounds, smells, noises. A psychology major friend of mine explained this to me. She recalled -- correctly, hopefully -- a study in which some subjects were presented with their real class picture from early childhood while someone related a false event that supposedly happened to them during childhood. Those who had the picture in front of them were more likely to later recall that event as real. Of course, some memories are easier to implant than others. An article from the UK Guardian describes one study in which "50% of volunteers were persuaded they had taken a ride in a hot-air balloon when they had not. But when Kathy Pezdek of the Claremont Graduate University, California, tried to make people believe they had received a rectal enema, she met with almost universal resistance."

I am no scientist, but this does reveal something about human cognition. It seems that, in layman's terms, some wires are getting crossed up there in the old noggin'. 

 Loftus' research is credible, or at least law enforcement and lawyers seem to think so. Her work is specifically part of the criminal and forensics studies field, and the Justice Department is interested. Obviously, her work questions the credibility of eyewitnesses to crime and might exonerate some who are accused of crimes in which repressed memories are used as evidence. Or, conversely, her work suggests that some people can be falsely convinced they committed crimes.   

Here is one example I found on the exploratorium website:

Consider, for instance, the experience of memory researcher Donald Thomson. Thomson appeared on a television show about the unreliability of eyewitness testimony. Not long after the show aired, he was picked up by the police and placed in a lineup. A distraught woman identified him as the rapist who had attacked her. 

 Thomson had an unshakable alibi--the rape had occurred when he was on the TV show. The victim had been watching Thomson on TV just before the rape, and had confused her memory of Thomson with her memory of the rapist. 

Here is a youtube video about Loftus' work.

And here is an activity you can try that explains the phenomenon, particularly the associative nature of cognition that can work against you!