In a giant leap into the future, researchers have developed a computer program that reacts to optical illusions the way humans do. Despite the tremendous growth in technology, scientists have not been able to create a robot that replicates the subconscious behavior of humans. In the wake of this development, scientists believe that future robots can actually have a brain similar to that of a human.
The reaction exhibited by human beings towards illusions is an inherently developed characteristic that filters and deciphers a complex surrounding. The way people perceive colors, shadows, contrasts in light, etc is a reflection of human behavior towards a seemingly illogical event. What ensues is the effort to understand an object based on its characteristics, like ambient light, shadows, etc. Over a period of trial and error attempts, babies grow up learning how to decipher their surroundings. And when the brain fails to perceive the surrounding in the right way, an optical illusion is created.
Beau Lotto and David Corney of University College London, UK, have succeeded in creating a program that perceives the surroundings exactly like humans do. It tries to predict the characteristics of an object or an image based on what it had learnt in the past. And like humans, it is not spared from optical illusions.
The program was tested on lightness illusions that usually throw the human brain off-track. Images of light objects on darker backgrounds, and vice versa, usually cause illusions in humans. The program had reacted to such images the way humans are prone to do, overestimating lighter shades and underestimating darker ones.
Typically, the image shown above had humans viewing grey areas as darker on a black stripe and lighter when on a white stripe. Computer programs created earlier exhibited a similarity in reaction only in either of these illusions. However, this program falls for both illusions.
The finding has certainly proved that computers can make mistakes and therefore, they can be closer to humans in their behavior than what we have been thinking so far!
Via: Newscientist
