An artificial brain has enabled new understandings on the manner a human brain is processing moving images, explaining some confusing optical illusions.
By utilizing decades' worth of data from human motion insight research, scientists have trained an artificial neural network to approximate both the image sequences' speed and direction.
This computer network, closely modeled on the part of the human brain is called MotionNet. It is designed to match what's described as the motion-process structures inside the human brain.
This has enabled the study authors to discover human visual processing features that can't be gauged directly in the brain.
This research, which Journal of Vision published, uses the artificial system to describe how space and time information is integrated in the brain to produce humans' perceptions or misperceptions of moving images.
'Phi' and 'Reverse-Phi' Motion
According to a EurekAlert report, the brain can easily be deceived. For example, if there is a black spot on the left side of a screen, which is disappearing while a block spot is appearing on the right, the spot is seen moving from left to right. Such a movement is called the 'phi' motion.
However, if the spot appearing on the right side is white on a dark background, the spot is seen moving from right to left. This movement is now called 'reverse-phi' motion.
The study authors replicated revers-hi motion in the MotionNet system and discovered that it made the same errors in perception like a human brain, although unlike the latter, they could look closely at the artificial system to find out the reason for such an occurrence.
What the researchers found was that neurons are 'tuned' to the movement's direction, and in MotionNet, 'reverse-phi' was stimulating neurons tuned to the direction opposite to the actual movement.
The Dots
The artificial system revealed too, information about this common illusion and that's the reverse-phi motion is impacted by how far the dots are apart, in the reverse to what would be expected.
Dots that move at a constant speed seem to move faster, as shown in Cambridge University's YouTube video below, at a shorter distance, and more slowly if with a longer distance.
Researcher and the study's first author, Reuben Rideaux from the University of Cambridge's Department of Psychology said they have known about reverse-phi motion for a long time, although the new model produced a totally new prediction about how it is experienced, which nobody has ever looked and tried before.
Perception of Motion
Humans are sensibly good at working out a moving object's speed and direction by simply looking at it. It is how one can catch a ball, calculate depth, or decide if it is safe to cross the road.
This, the study specified, is done by processing the light's changing patterns into a 'perception of motion,' although many aspects of how it takes place are still not understood.
Rideaux explained, it is quite difficult to directly gauge what is happening inside the human brain when motion is perceived that even the best medical technology cannot show the whole system at work. With MotionNet, the first author said, they now have complete access.
RELATED ARTICLE: Research is a Step Closer to Brain-Machine Interface Autonomy
Check out more news and information on Human Brain on Science Times.
