Researchers look at how we understand visual images

Aug 9 2005

The figure is famous: a deceptively simple line drawing that at first glance resembles a vase and, at the next, a pair of human faces in profile. When you look at this figure, your brain must rapidly decide what the various lines denote. Are they the outlines of the vase or the borders of two faces? How does your brain decide?

It does so in a fraction of a second via special nerve circuits in the brain's visual center that automatically organize information into a "whole" even as an individual's gaze and attention are focused on only one part, according to Johns Hopkins researchers writing in a recent issue of the journal Neuron.

"Our paper answers the century-old question of the basis of subconscious processes in visual perception, specifically, the phenomenon of figure-ground organization," said Rudiger von der Heydt, a professor in the Zanvyl Krieger Mind-Brain Institute. "Early in the 20th century, the Gestalt psychologists postulated the existence of mechanisms that process visual information automatically and independently of what we know, think or expect. Since then, there has always been the question as to whether these mechanisms actually exist. They do. Our work suggests that the system continuously organizes the whole scene, even though we usually are attending only to a small part of it."

The report, based on recordings of nerve cells in the visual cortex of macaque monkeys, suggests that this automatic processing of images is repeated each time an individual looks at something new, usually three to four times per second. What's more, the brain provides what von der Heydt calls "a sophisticated program" to select and process the information that is relevant at any given moment.

"The result of this organization is an internal data structure, quite similar to a database, that allows the attention mechanism to work efficiently," von der Heydt said. "An image can be compared with a bag of thousands of little Lego blocks in chaotic order. To pay attention to an object in space, the visual system first has to arrange this bag of blocks into useful 'chunks' and provide threads by which one or the other chunk can be pulled out for further processing."

He noted that the research provides the theoretical foundation that might one day lead to better diagnosis and treatment of human brain disorders.

"The last decades have seen rapid progress in the neurosciences at a very broad front, particularly at the molecular and cellular levels, and this progress makes it increasingly clear that we still lack sufficient understanding of brain function at the 'system level,'" he said. "We need to understand the basis of mental processes. Single cell recording in animals is only one approach to this formidable task. It is complemented by new brain imaging techniques, traditional psychophysics, psychology and computational and theoretical neuroscience. … Understanding the function of the visual cortex will help to interpret neurological symptoms in diseases that produce disorders of vision."