CU fMRI Research Explores Brain

A monkey sits in a pitch-black room waiting for a tiny point of light to appear in front of him. When it does, he stares at it intently until it blinks off again. His reward is a sip of delicious juice from a tube.

It may be a just a game to the monkey, but for Columbia scientists Yevgeniy Sirotin and Aniruddha Das, it is an experiment. In fact, although Sirotin and Das planned to use this particular experiment only as a baseline for further research, it has unexpectedly produced results that will change the way many images of the human brain are interpreted.

The discovery has to do with a technique called fMRI, or functional magnetic resonance imaging, which detects local changes in blood flow in the brain. Until now, most scientists thought that the blood flow in a given region of the brain would allow them to directly calculate the amount of neuronal activity.

This assumption was hardly a guess. Although the experiments required to test it definitively were too invasive to perform on humans, a comprehensive study of anesthetized monkeys in 2001, as well as subsequent observations in conscious animals, seemed to confirm that blood flow and neuronal activity were indeed closely linked. But as Sirotin and Das prepared to embark on a complex study of visual perception, they wanted to make sure that they started from the basics.

They designed an experiment in which one of their two monkeys would sit and wait for a small light, which would appear for a few seconds at regular intervals. They trained the monkey to fixate his gaze on this light. While the monkey was absorbed in his task, the scientists used a special camera to peer into his brain through a small window in his skull. The camera measured the ebb and flow of blood in the brain’s V1 cortex, a specific area which handles the first stages of visual processing. They had also implanted tiny electrodes in this same region, which detected the electrical activity of nerve cells in the area. In this way, they could directly measure both blood flow and neuronal activity.

When Das and Sirotin showed the monkey a visual stimulus—such as a picture or video—at the same time as he was fixating on the light, both blood flow and neuronal activity increased in unison, as expected. But when the researchers didn’t show the monkey a visual stimulus, something strange happened.

“It was really a serendipitous finding,” Sirotin said. The researchers observed that, in the absence of a visual stimulus, the nerve cells in the V1 cortex did not change their activity from that of the dark-room baseline. But the blood flow to the region kept rising and falling in the same cycle as before, as if the brain were able to predict which of its parts would be needing blood flow soon. Such anticipatory blood flows had never before been observed in any animal.

These findings have implications for the interpretation of human fMRI images, which detect only blood flow and not neuronal activity.
According to Sirotin, the results show that it is possible to predict blood flow given knowledge of neuronal activity, but not the reverse.

Does this mean that the results of thousands of prior fMRI studies must be discarded? Not necessarily. “It’s really a question of what sort of conclusions you can draw,” Sirotin said. He added that while some prior studies would have to be reevaluated, most researchers have been careful to note that the blood flows detected on fMRI are not exact proxies for underlying neuronal activity, so that their results can still be trusted.

Sirotin and Das have published their results in the prestigious journal, Nature. And, as is frequently the case in science, the work raises more questions than answers. How does the brain organize the anticipatory blood flows? Does an anticipatory rush of blood to a certain brain region actually improve its performance?

The latter question is of particular interest to Justine Ordinario, BC ’09, an undergraduate in Das’s lab. To help answer it, she has been training the monkeys to perform tasks for new experiments, which is no easy job.

“They’re pretty smart animals,” Ordinario said. “So sometimes they come up with different strategies for doing different tasks. So we have to figure out what their strategy is to make sure they’re not doing what we don’t want them to do.”