In The Field

SfN 2009: Why fMRI is still useful

Functional MRI has been getting a bad rap lately, with recent papers and posters critical of fMRI analyses receiving a frenzy of media attention. These have generated a harsh reaction from the public; many journalist friends of mine have declared they’ll never write about an fMRI study ever again.

Sure, some people are seduced by the pretty pictures and like to think they let scientists “see our thoughts”, and this fantasy should be nipped in the bud. But while fMRI has its limitations, it’s a valuable technique that, when performed correctly, can give neuroscientists a unique insight into the brain.

SfN gave me a nice opportunity to get an fMRI veteran’s perspective on the state of fMRI in science and the public. John Gabrieli at MIT has been working in the field since the early years (fMRI only came about in the early 90’s), and has written and spoken extensively on the power and limitations of the technique. Here he participated in the “Meet the Experts” series, where he gave a presentation about the technique to a small group of mostly post-docs and students over breakfast.

He noted that the flashy colorful pictures of brain activation are statistical maps — not actual changes in blood flow. For example, if you have a very powerful visual stimulus, like a flashing light, the signal in the visual cortex would change by about 3%. Same for the BOLD signal in the motor cortex when you wave your arms around wildly. But for most interesting activities — thoughts, feelings, desires, memories — a signal change of half of 1% would be “a really good day”, he says. He points out that if we had been doing imaging before the famous patient H.M., we would have no idea the hippocampus was involved in memory formation; “we would have spent 20 years thinking the hippocampus was not involved in memory, because in dozens of memory tasks we couldn’t get hippocampus.”

Gabrieli says this super-weak signal means we’re actually missing a lot of stuff, and are at a far greater risk for false negatives than we are for false positives. People only get into trouble with false positives when they try to overcompensate for this low signal, and Gabrieli detailed ways to avoid this trap.

Gabrieli also spoke about some of the newer uses for fMRI, beyond just trying to see “where” things happen in the brain. For example, if scientists could find brain activation patterns in young people that were associated with the later development of diseases like schizophrenia or depression, they could use fMRI as a far more accurate predictor of psychiatric disease than genetic testing, which is extremely limited.

Another hot area of fMRI research is “real time” fMRI, in which a subject’s fMRI signal feeds back to what they see. For example, if a certain area of the brain is active when the subjects are ready to remember a picture, the fMRI can wait until this brain region is active before presenting the picture. A few people have already started thinking about clinical applications of “real time” fMRI, such as enhancing learning, behavioral therapy and giving people “control” over their auditory cortex to stop tinnitus.


Comments are closed.