False Signals Cause Misleading Brain Scans

A functional magnetic resonance image of the brain. i i

This functional magnetic resonance imaging, or fMRI, of a brain shows activity in the ventral tegmental area (VTA). Scientists gave brain scans to people who said they were in love, while showing them photographs either of their sweeties or other acquaintances, and found that the subjects' VTAs had increased activity when thinking about their loved one. New research says these images may be misleading. Lucy Brown/AP/Albert Einstein College of Medicine hide caption

itoggle caption Lucy Brown/AP/Albert Einstein College of Medicine
A functional magnetic resonance image of the brain.

This functional magnetic resonance imaging, or fMRI, of a brain shows activity in the ventral tegmental area (VTA). Scientists gave brain scans to people who said they were in love, while showing them photographs either of their sweeties or other acquaintances, and found that the subjects' VTAs had increased activity when thinking about their loved one. New research says these images may be misleading.

Lucy Brown/AP/Albert Einstein College of Medicine

Read The Original Research

You can download the original paper here.

Love lights up one area of the brain; rejection lights up another.

At least that's what it looks like in images produced by the latest brain scanners. The pictures typically show a dark cross-section of the human head with vivid bursts of red or yellow in places where the brain seems to be responding to something.

The images appear amazingly crisp and precise. But scientists say the truth behind them is a little fuzzier.

"These are difficult, challenging experiments," says Chris Baker, chief of the Unit on Learning and Plasticity in the Laboratory of Brain and Cognition at the National Institute of Mental Health. And the images they produce can be misleading, he says.

Static Interference

Most of the pretty pictures come from a technology called functional magnetic resonance imaging, or fMRI. It detects tiny changes in the amount of oxygen carried by blood to brain cells. More oxygen generally means more brain activity.

But it takes a whole lot of computer processing and human judgment to get from oxygen levels to a snapshot of love in the brain.

The challenges begin the second a researcher slides someone into a scanner and starts looking for signals that suggest increased brain activity.

"The problem with functional imaging is that the signals we're trying to get at are quite weak, and there's a lot of noise," Baker says.

The "noise" is in the form of false signals. These can come from the scanning equipment itself, but a lot of it comes from the person being scanned.

Every heartbeat affects the flow of blood, which changes the signal. Every tiny head movement blurs the image.

Thinking About Something Else

And then there's noise from all the stuff your brain is thinking about that has nothing to do with the experiment.

"It's like listening to a radio station on which there's a lot of static," Baker says. "So sometimes the noise will obscure being able to hear the people or being able to pick up the station at all."

And an fMRI study isn't as simple as listening to just one faint radio station from one place in the brain.

Scientists use fMRI to create three-dimensional maps of activity throughout the brain. A typical scan gathers data from 100,000 different places, and it's up to the researcher to decide which spots actually have something to say.

"You're tuning the radio, looking across different frequencies," Baker says. "And for each frequency you go to, you're trying to pick up and see, was there really somebody speaking there? Did I hear a song at that particular point in time?"

And if there was a song, was it about love? Or rejection? Or nothing at all?

Baker says the signals in fMRI studies are often so weak that researchers have to stimulate a person's brain over and over again to see any pattern at all.

For example, let's say you want to know which areas of the brain respond to faces, something Baker has investigated.

"Typically, we're going to have to present lots and lots of faces" to be able to detect an increase in brain activity in a particular area, he says.

Too Good To Be True

Then, you have to repeat the experiment on 15 or 20 more people to see if the same areas light up in everyone. In many cases, it won't.

And face recognition is a pretty simple brain process compared with experiencing an emotion like love, or telling a lie. That's why scientists are wary of scans that claim to identify these things.

The fMRI studies frequently produce billions of data points — many of which represent static, not a true signal.

Baker says that when you're searching an ocean of noisy data, it's possible to find patterns produced purely by chance.

"There are situations in which you could apparently produce a signal out of just noise," he says.

Baker doesn't think that happens very often, but he showed how it could in a recent paper published in Nature Neuroscience.

That paper is one of two published this year that suggest the results of many fMRI studies aren't as robust as they appear.

Hal Pashler, a professor of psychology and cognitive science at the University of California, San Diego, is an author of the other paper. It was published in the journal Perspectives on Psychological Science.

Pashler says he and his co-authors wrote the paper because they were seeing results from fMRI studies that seemed just too good to be true.

The team began to ask why and zeroed in on a statistical practice sometimes called "double dipping," a kind of circular reasoning.

The easiest way to understand double dipping is to imagine something that has nothing to do with brain imaging, Pashler says.

"Suppose you wanted to figure out whether anyone in a town was able to predict the temperature three weeks from now," he says. You'd have everybody in town make a prediction, then come back three weeks later and see how they did.

There might be 50 or 100 people whose predictions were spot on. You might conclude that all of these people are just brilliant at predicting the temperature.

But that's probably not the case because some of the people probably got it right purely by chance, Pashler says.

"So if you test them again," he says, "you're going to find that they'll do better than most people in town, but they won't do as well as you think they did based on the data collected when you first picked them out."

Not As Strong As They Seem

Pashler and Baker say many fMRI studies do something a lot like this when they dip into the same data twice: once to pick out which parts of the brain are responding, and again to measure how strong the response is.

The scientists say that's probably made a lot of results appear stronger than they really are.

The papers criticizing fMRI studies have been the subject of furious debate in the world of brain science.

But even the strongest defenders of fMRI acknowledge that there are problems, and that there's a need to exercise great caution when transforming messy data into tidy images.

And it's not just the unwary public misled by overstated results, Pashler says. Funding agencies and other researchers are also influenced.

That could mean some areas of the brain are getting attention they don't deserve, while other potentially important areas are being ignored.

All because of images that aren't as clear as they seem.

Comments

 

Please keep your community civil. All comments must follow the NPR.org Community rules and terms of use, and will be moderated prior to posting. NPR reserves the right to use the comments we receive, in whole or in part, and to use the commenter's name and location, in any medium. See also the Terms of Use, Privacy Policy and Community FAQ.