Irrational eyes: seeing curves that aren't there

We can't always believe our eyes - we've all experienced visual illusions where our brains create visions of objects that aren't actually there. Why this happens is often explained in terms of our brains filling in the gaps logically: we see what is most likely to be there. However, Professor Bart Anderson, from the School of Psychology at the University of Sydney, has discovered new forms of visual illusions which cause the brain to create highly improbable shapes that aren't there.

In research published in the international journal Current Biology in March 2011, Professor Anderson, along with Judit O'Vari, a PhD student in the School of Psychology at the University of Sydney, and Assistant Professor Hilary Barth, from Wesleyan University, USA, presents new moving displays that make us see unnecessary contours.

The new findings challenge models that explain visual perceptions in terms of our brains 'seeing' the most optimal and probable visuals.

"If you look at these moving shapes, you see a really vivid illusion, in addition to the shapes which are really in front of you. For example, in the first animation presented here, we see a rectangular outline in white, with four solid grey circles moving towards and away from the rectangle, appearing to cover parts of the rectangle as they move. We also see a wave-like line that appears to spread outwards through each circle as it covers the rectangle, which seem to be connected between the four circles to create a roughly oval wave moving inwards and outwards," explained Professor Anderson.

Professor Bart Anderson and team have found a new form of visual illusion which make us see unnecessary contours - we see a wave-like line that appears to spread outwards through each circle as it covers the rectangle. In this version, there is low contrast between the moving circles and the background, so we see a strong visual illusion.

"This pulsating 'wave' doesn't exist at all - when we look at the circles we can see that no such wave lines exist, but we perceive them anyway.

"When we add the smaller pale circles moving inwards and outwards over the rectangle and solid circles, this wave illusion becomes even stronger - we actually join the four pale circles to create something like a rubber band being stretched in and out at the points of the dots.

In this version of the display, there is medium contrast between the moving circles and the background, so we see the visual illusion less vividly than in the low contrast display above.

"These simple occlusion sequences - where the circles cover the rectangle - elicit vivid imaginary lines that aren't necessary to make sense of the moving sequences," explained Professor Anderson.

In the natural world, objects which have some of their boundaries hidden by other objects are common. Our visual systems have evolved to make sense of the partially covered shapes surrounding us, but our brains don't necessarily create the most logical or probable perceptions.

"The fact that we 'see' these unnecessary extra lines, challenges models which explain visual interpolation in terms of our brains rationally coming to the most logical conclusion," said Professor Anderson.

"Our new findings suggest that in explaining how our visual system 'fills in the gaps' when looking at interrupted or partially obscured shapes, we need to consider a broader space of computational problems and implementation level constraints to understand how we end up seeing what we do."

The team conducted experiments to test which factors modulate the vividness of the illusion, to see whether the brightness of the circles or the rectangle had an effect on how strong an illusion was created.

"We found that people saw a stronger illusory line when there was not much contrast between the brightness of the moving circles and the background, but they saw a weaker illusory line when there was not much contrast between the brightness of the rectangle and the background," said Professor Anderson.

"So how vividly we see the illusion depends on the relative edge contrast of the circles, which do the covering, and the rectangle, which becomes covered. So it looks like how strongly we see this illusion is based on a relative measure of edge strength between the occluded contour and occluding surface."

"When we add the smaller dots, we found that there is the same effect with the contrast - you get a stronger illusion when there's not much contrast between the larger circles and the background, or when there's high contrast between the rectangle and the background. This suggests that the moving dots do not cause the illusion, but just reveal its presence."

The team found the illusion is extremely robust - it can be observed over a broad range of contrasts, contrast polarities, speeds, and shapes of both the occluded and occluding figure.

"The main puzzle is to understand why the visual system constructs these additional illusory surfaces, when there are clearly visible surfaces that account for all of the image data," said Professor Anderson.

"Any model to explain what's happening when we see these illusions will have to consider a broader space of computational problems for which these illusions emerge as adaptive by-products or consider implementation level constraints that prevent the system from achieving optimal forms of inference when we look at displays like this."

Read the journal paper in Current Biology

Contact: Katynna Gill

Phone: 02 9351 6997

Email: 2d1b2308252825760d24203408383f0e363d03495537254b033c