Maunsell and his colleagues at Baylor College of Medicine have worked with monkeys trained to fix their gaze on a central spot on a computer screen and then—without moving their eyes—shift attention to other targets. Meanwhile, a computer-aided sensing system records the electrical activity of the neurons in the brain that are receiving stimuli from the retinal cells that capture the object of the animal's attention.

This research has shown that when monkeys thus shift their attention, a surge of electrical activity occurs in those neurons. The investigators have demonstrated such an effect in many areas of the brain, in nerve cells specialized for different features such as detecting edges and motion as well as recognizing patterns.

More recently, the researchers changed the test conditions. The monkeys were trained to concentrate on a single dot. When dedicated neurons detected the target dot, their electrical activity spiked to twice the normal firing rate, and the same cells "turned down" their response when they encountered a similarly moving dot that wasn't the one on which they were trained to focus.

"The behavioral result is that you get improved perception or faster reaction times when the monkeys detect a small change or respond to it," Maunsell says. "What attention is doing is just altering the sensory representation the animal will use to make his decisions."

Maunsell emphasizes, however, that the allocation of attention is a dynamic, constantly changing process, and the strength of responses in the brain cells "can fluctuate over a fraction of a second as the animal directs more or less attention to different parts of the visual scene."

Maunsell is currently conducting experiments to discover how the brain translates a visual image's information into a motor response. So far, it looks as if this process is based on information from a limited voting body, so to speak, rather than a large population. A relatively small number of neurons—hundreds, perhaps—are involved in different areas of the cortex.

This kind of research, Maunsell says, reflects a new stage in the daunting journey to understanding the workings of the brain. "I view the last 30 years as coming to grips with how things are laid out in the brain and where visual images are represented," he says. "We have a decent first draft."

Now, he says, researchers are delving into the still-mysterious processes "by which, using the 1.2 billion neurons in the visual cortex, the important bits of information are extracted and an appropriate motor response is determined."

We've all seen puzzling photos of objects that are unrecognizable until we're told or eventually figure out that they are small parts of something larger—an architectural detail, perhaps, of a familiar building. What was initially lacking was a context for the otherwise meaningless shape we were looking at. As soon as the context became apparent, recognition was a snap.

In one area of his wide-ranging vision research, Thomas D. Albright, an HHMI investigator at the Salk Institute for Biological Studies, in La Jolla, California, has been studying the crucial importance of contextual clues to visual perception. Context can mean many things, from the physical features of a visual target's environment to memories stored in the brain that are associated with the object. Context, says Albright, helps us "recover" information that's missing from the original image captured on our light-sensitive retinas.

At the first of several stages of increasingly sophisticated processing and interpretation, the retina senses mainly a patchwork of light, dark, and color—contrasts without recognizable shape or significance. In the next round of processing, Albright explains, brain cells that respond exclusively to certain features of an image begin providing rough interpretations of the visual scene. Some specialized cells are activated when they sense an edge or a contour, others when they detect motion in a specific direction, and still others when certain colors or brightness levels are present.

	PAGE 5 OF 6	Go Back | Continue