Mailing List

Contact Us

Staff

	Eric Schwartz, a CAS professor in the department of cognitive and neural systems, is using this functional magnetic resonance scanner at the Massachusetts General Hospital’s research facility to peer into the visual cortex. The real-time images are shedding light on how the brain makes sense of the visual world. Photo by Bruce Caplan

By Tim Stoddard

It’s sort of like looking down a hall of mirrors: trying to see how the brain sees while it’s, well, seeing. That’s what Eric Schwartz, a CAS professor in the department of cognitive and neural systems, is doing, harnessing the power of functional magnetic resonance imaging (fMRI) to watch the visual cortex in action. The results may soon not only shed light on how the brain makes sense of the visual world, but also help improve the power and accuracy of fMRI technology.

When light lands on the velvety surface of the retina, specialized cells go to work like a legion of telegraph operators, encrypting the image as electrical impulses that zip along to the primary visual cortex, called V1 by researchers. Neuroscientists have known for 50 years that the brain arranges these signals in a kind of representational map, with each point on the retina routed to a unique point in V1. Researchers think this map is then copied onto 30 to 40 different areas within the visual cortex, each of which interprets a different aspect of the image, such as colors, edges, and motion. But with his colleagues at CNS, Schwartz has done recent computational work that suggests the structure of the brain is simpler and more elegant, processing images in only a few compartments.

“ This model would change the whole concept of how the cortex is organized,” Schwartz says. “It’s as if in astronomy you didn’t have a concept of a galaxy, but only had one for a solar system. And then somebody came in and said, ‘Well, actually there’s a higher level of structure that you didn’t see before.’”

With a $2 million grant from the National Institute for Biomedical Imaging and Bioengineering, a new branch of the National Institutes of Health, Schwartz and Bruce Fischl (GRS’97), an assistant professor of radiology at Harvard Medical School, are using fMRI to investigate the primary visual cortex in unprecedented detail.

Magnetic vision

In the 1960s neuroscientists began using implanted electrodes to study nerve activity in the visual cortex of macaque monkeys. In the standard experiment, the monkeys were shown a bull’s-eye pattern of white lines on a black background. The researchers hoped to find a mathematical function describing how the brain takes in the optical information and distributes it within the primary visual cortex, but they found no clear patterns.

As a postdoc in 1976, Schwartz revisited their data and found that a complex logarithm describes the visual mapping fairly well. It turns out to be the same function for calculating the geometry of a magnetic field. When you drop iron filings over a bar magnet, they spread out in an hourglass shape along the axis of the magnet. “The mathematical function that plots out the retina-to-cortex mapping is the same as that which tells you how to plot out the field lines of a magnet,” Schwartz says. “In some sense, the physical world is following the simplest possible idea, and so is your brain.” He emphasizes, however, that there are no bar magnets in our brains. “It’s the same mathematics,” he says, “but it isn’t the same phenomenon.”

With CNS colleagues, Schwartz recently did a detailed analysis of several topographic maps described by other researchers, and found that they have almost the same mathematical structure as V1. Their models suggested that instead of 40 distinct topographic maps, there were only 2 or 3 separate “super-maps,” with many submaps nested within them like Russian dolls.

Hardware tweaks

While this is still theoretical work, Schwartz says, it gives researchers a new tool for understanding fMRI. First developed about a decade ago, fMRI works in real time by measuring oxygen levels in the blood throughout the brain, indicating which neurons are getting busy and consuming oxygen. But fMRI and MRI are only about 95 percent accurate; there are minor distortions in the images that are trivial in most clinical applications but problematic for precise measurements.

Schwartz plans to use the mathematical model he’s developed as a benchmark, and compare it to what he actually finds on the fMRI scans, using ever more powerful fMRIs at a Massachusetts General Hospital research facility. “You can’t really fix the imaging if you don’t have a benchmark for what correctness is,” he says.

What kind of biomedical progress can we expect from an even more sophisticated MRI? “The consequences of improving imaging quality are hard to predict,” Schwartz says. “In the history of science, pushing the measurements further and further has been where progress comes from, and I think the instrumentation drives the science. When it gets better, the science gets better.”

12 September 2003
Boston University
Office of University Relations