Thursday, March 18, 2010


I've done a lot of reports on false memories and how memories are formed through cognitive processes and in the brain. One thing that I've learned over and over again is that our memories are often not as good as we think they are, and that there have been many court cases where eye witness testimonies were turned over by physical evidence. In the past 10 years eye witness testimonies have actually come under much greater scrutiny, to the point that there have been discussions of discounting them as evidence in a court of law. One example of how eye witness testimonies can be misconstrued is the common practice of having a witness pick the perpetrator from a line-up of individuals, or the sometimes used alternative of giving the victim a book of pictures of criminals and asking them to pick out the person who wronged them.

The problem here is that there are common elements to the human face that can sometimes be confused, along with the often felt pressure that one must pick SOMEONE from the line-up. What they have found is that if presented with several people a victim will often pick the person CLOSEST to the perpetrator, even if they barely resemble them. Giving them a book of mugshots is even worse because by the 20th mugshot they have so many facial features in their mind that whatever picture they had of the perpetrator is usually all but completely gone.

Then this idea just popped into my head. Why not take one facial feature, such as eyes, and then have seven different faces that are the same except for the eyes that are different. Then have another set of seven pictures that are the same except for the mouth that is different. Continue with sets of pictures going through many different facial features. With each set ask the person to identify who looks the most like the perpetrator. Also, to keep them from over-thinking it or getting too many facial features stuck in their head and confusing them you could show each picture for only 1-2 seconds and then remove it and ask them to go with their gut (human's ability to recognize pictures is extremely fast, so doing this is actually reasonable). Then, after you have gone through several sets combine the facial features they chose into a face. There would be several possible combinations, so maybe come up with 7 faces that utilize the features they gave you and then ask them to identify which one is most like their perpetrator. Lastly, take the pictures of individuals that you wanted the victim/witness to choose from and run them through and the computer will tell you which one is most like the individual that the victim described.

I want to talk to a Cognitive Psychology professor here at school and see if this is a reasonable idea considering the cognitive processes of human beings. I know that all the technology for doing such a thing has already been invented, but I wonder if it would really overcome the challenges that have come to be associated with face recognition that victims and witnesses often have.

Hmmm...just a random thought of Nathan Good.......


  1. fascinating idea, really. Have you read Blink by Malcolm Gladwell?

  2. I have not read that book, although I have heard of it.