Cystal [eye]ball: Precognitive Eyes are Normal

News

Not a member yet? Register for full benefits!

So many of our everyday tasks, like catching a thrown object, manoeuvring through a crowded lobby, blasting an alien invading force to smithereens - all rely on visual perception. They rely on the ability to perceive a moving environment, to process it and respond in timely fashion. If a cricket ball is speeding towards you, it is no good watching it come in, and not moving. That leads to pain.

It takes our brain nearly one-tenth of a second to translate the light that hits our retina into a visual perception of the world around us. Whilst that might not seem like much, recall the last time you navigated a hectic virtual environment, say a combat-orientated gameworld, where the enemy was closing in, and you only had a frames per second of ten or less. In that situation, its safe to say you died, a lot. Objects flying towards you at any velocity, from a simple toss, to an inbound rocket, can all move a great distance in a tenth of a second.

Restricting yourself to processed data that was a tenth of a second old, is near suicidal in a high-octane virtual life. In the physical world, there would be nothing near about it.

Because of the 10th of a second delay, Assistant Professor of Cognitive Science at Rensselaer Polytechnic Institute Mark Changizi has developed a case arguing that our visual system needs to do more than just report current events - it needs to see the future as well.

As it takes our brains a tenth of a second to analyse data, he argues that our eyes and optic nerves are delivering data that is predicting the environment ten seconds into the future. What's more, he has found proof to back his theory up.

The Hering illusion above, creates the perception that the straight lines curve increasingly, the closer to the perceived vanishing point they come.

The optical illusion occurs because our brains are predicting the way the underlying scene would project in the next moment if we were moving in the direction of the vanishing point.

Changizi claims the visual system has evolved to compensate for neural delays, allowing it to generate perceptions of what will occur one-tenth of a second into the future, so that when an observer actually perceives something, it is the present rather than what happened one-tenth of a second ago. Using his hypothesis, called "perceiving-the-present," he was able to systematically organise and explain more than 50 types of visual illusions.

The illusions, which have been enthusiastically discovered and documented over the years, have never been organised into trends - no attempt has been made to discover the root causes, and whether or not they have emergent similarities that would categorise the various effects produced.

"Illusions occur when our brains attempt to perceive the future, and those perceptions don't match reality. There has been great success at discovering and documenting countless visual illusions. There has been considerably less success in organising them," says Changizi, who is lead author on the paper. "My research focused on systematising these known incidents of failed future seeing into a 'periodic table' of illusion classes that can predict a broad pattern of the illusions we might be subject to."

The Hering illusion, depicted on the left, is one such. A spoked wheel with six vertical lines drawn across it, spaced equally such that three are on either side of the wheel centre, with the centre equidistant between them.

Although the lines are straight, they seem to bow out away from the vanishing point. The optical illusion occurs because our brains are predicting the way the underlying scene would project in the next moment if we were moving in the direction of the vanishing point.

In other words, they are bowing out like a doorframe.

"Evolution has seen to it that geometric drawings like this elicit in us premonitions of the near future," says Changizi. "The converging lines toward a vanishing point are cues that trick our brains into thinking we are moving forward-as we would in the real world, where the door frame seems to bow out as we move through it-and we try to perceive what that world will look like in the next instant."

In the Orbison geometric illusion, the squares closer to the center of the image seem larger than those on the outside of the drawing.

Again, we have evolved our way of perceiving such that our brains are predicting the 'next point' and how we would perceive this as a 3D construct, if we were moving towards it at a walking pace, over a tenth of a second.

Both illusions here are geometric in nature (revolving around a vanishing point), but different types of optical illusion abound by the thousands. Seen every day in the modern world. We experience countless illusions in our lifetime.

Beyond geometric, Changizi was able to identify 27 other classes of illusions. He organised them into 28 predictable categories classified on a matrix that distributes them among four columns based on the type of visual feature that is misperceived (size, speed, luminance, and distance) and among seven rows based on the different optical features that occur when an observer is moving forward.

He then culled hundreds of previously documented illusions to test whether they would follow the appropriate prediction as determined by the table. His discoveries were that each and every illusion had a place in the table, laid out in increasingly obvious patterns as each was eliminated - they derive from an extrapolation of future events.

The results of this study have implications for a cross section of fields.

Neuroscientists have evidence to work with that the optic system is doing more than simply sending information to the brain to process - the eyes and optic nerve are preprocessing the images as well.

Artificial eye design directly benefirts from understanding this phenomenon, as to does the design of display systems, and interfaces, both of whom can now start to assemble a body of work based on predication in vision which has implications for interaction with interface systems, and quite possibly with ways to visually send more information with less data - one of the holy grails of visual virtual environment work.