In 1880 an Italian peasant named Bertino survived a horrific accident that cracked open his skull and left sections of his brain exposed. Surprisingly Bertino felt fine, even though one could see blood pulsating through his frontal lobes. His physician, Angelo Mosso, noticed something else very strange. Every time church bells rang in town, blood surged through Bertino’s lobes. Mosso took a guess that the blood surged because the bells reminded Bertino of prayer. When Mosso asked Bertino directly, “Do the bells make you think of prayer?” Bertino answered, “Yes.” At that moment blood again engorged the exposed veins. Then Mosso asked, “What is eight by 12?” Bertino answered, “96.” More blood pulsed through. The link Mosso had stumbled upon was perhaps the first connection made between blood flow and brain activity — a serendipitous connection that, more than century later, would become the foundation for a revolutionary tool to study the brain: functional magnetic resonance imaging, or fMRI.
Today, fMRI is routinely used to turn our dull grey matter into technicolor images of reds and yellows, which can tell us where in the brain we experience intangible emotions like romantic love, political passion, even petty glee over others’ misfortune. This technological mind reader has captured the imagination of both scientists and the public. Indeed, scientific journals are devoting more and more pages to what many scientists call one of the greatest scientific advances of the last quarter century.
“This is the first time in history we have a non-invasive, high-resolution, totally safe way to look at a human brain,” says Joy Hirsch, director of the fMRI Research Center at Columbia University. “To look at how the brain drives and controls behavior — this is the single most important question in neuroscience.”
But many caution that the flurry of excitement may be premature. Some worry that the dazzling brain images are usurping other important areas of psychology and convincing laypeople that science can do more than it can. Already the commercialization of fMRI is in the works. Private companies like No Lie fMRI, Inc. and Cephos Corporation are launching to capitalize on the potential for fMRI to replace the polygraph in lie detection. But the very scientists whose research has spawned these radical uses say it is too early to start depending on fMRI in ways that could have serious outcomes for an individual’s life. fMRI research is evolving and yet to be perfected, they say.
“There is a lot more work that needs to be done before I would try to commercialize it,” says APS Fellow and Charter Member (as well as APS President-Elect) John Cacioppo, Director of University of Chicago Center for Cognitive and Social Neuroscience. “We might predict someone is lying, but when we can only get 85 percent accuracy, that’s just not high enough.”
NUTS AND BOLTS
To be sure fMRI has starkly surpassed early attempts to see the brain in action. One of the first human neuroimaging techniques was a painful procedure called pneumoencephalography. Used in the early 1900s, the technology required that physicians drain the fluid surrounding the brain and then inject air in its place through holes in the skull, so the brain could be x-rayed. This crude technique remained state of the art until 1973, when CAT scanning (computerized axial tomography) yielded brain images that were detailed enough for diagnosis of brain abnormalities.
CAT scans were followed by positron emission tomography or PET scans, which measured the decay of radioactive chemicals in brain tissue, and also MRI, which detected magnetic force. But all of these early images were static. The “movies” of brain function only became possible with confirmation of the link between blood flow and brain activity. This opened the possibility of using MRI to study more than just brain structure. With this connection the active functioning of the brain could be predicted. Hence the lower case “f” added to MRI.
The fMRI technology is possible solely because of two fortuitous quirks of nature. When a certain area of the brain is active, it pulls more oxygenated blood to that area than is actually needed. No one knows why it overcompensates, but the result is a surfeit of oxygenated blood associated with increases in neuronal activity. Here is the second quirk: deoxygenated blood has magnetic properties (because oxygen neutralizes the effect of iron in the blood). So the ratio of oxygenated blood to deoxygenated blood can be picked up as a signal by the magnetic field of fMRI.
Experiments using fMRI take about 1 to 2 hours per participant and each scan costs approximately $1500. Subjects lie down on a narrow plank, within a tube, and remain as still as possible. Even a millimeter of movement can ruin the data. Depending on the study, subjects may be shown images of lines, asked to listen for the barking of dogs, or be engulfed in the smell of bananas.
The magnet picks up increases in oxygenated blood, and computers create images from the collected data – with reds and yellows marking the hotspots of neuronal activity. This is what scientists are referring to when they talk about a brain area “lighting up” when someone is looking at a photo of say, Jennifer Aniston. So according to critics, what fMRI really measures is a correlation between blood flow and activity, and not the brain activity itself.
But much of scientific progress starts with seeing some kind of correlation that begs explanation, like Angelo Mosso noticing surges of blood in the unfortunate Bertino’s brain. The data from fMRI provides another dependent variable that psychologists can use to help weigh competing psychological theories.
“It’s got meaningful information,” says Joshua Greene, assistant professor of psychology, Harvard University. “It’s not just swooshing blood everywhere – it’s far too specific and predictive of behavior.”
The most well known shortcoming of fMRI is its slow timing. The blood flow response takes about two seconds, but a thought can happen in milliseconds. So it’s difficult to say that a rush of blood is associated with a specific activity in the brain.
As a possible solution to this temporal lag, a few labs have started combining fMRI with other tools, like EEG (electroencephalography) and MEG (magnetic encephalograpy). EEG and MEG can provide millisecond-by-millisecond recordings of electrical brain activity. Joy Hirsch believes that the next quantum leap in neuroimaging will require this sort of fusion of imaging tools.
Right now her lab is experimenting with performing EEG within the fMRI scanner itself. Some investigators think it’s a waste of time, but Hirsh says there are strategies that can work. One of the difficulties is that the EEG signal is washed away by the giant magnetic field of fMRI, so one strategy is to protect the EEG from the magnet. “We use a technology that allows for the interweaving of the EEG in the gradients of the scanner,” she says.
This technique is proving beneficial in guiding surgeons who want to cut through exact areas of the brain that are prone to seizures.
“We can mark it in time and space,” says Hirsch. “And this helps the surgeon take out smaller parts of the brain and be more detailed in their work.”
Perhaps the most popular criticism of fMRI, and one fueled by the surge of media attention, is the tendency for fMRI studies to plant flags in different regions of the brain thereby claiming specific cortical zones that house concepts like love, self-will, and even the so-called “God spot.”
If we are not careful, says Marcus Raichle, considered to be one of the world’s experts on brain imaging, the technique might be viewed as modern day phrenology, (the theory developed in 1800 that claimed personality traits could be localized by bumps on the head). Already, the psychologist William Uttal wrote a critique of fMRI in his recent book called, “The New Phrenology.”
Certainly when we think about it there is more to religious belief than a specific neural center in the temporal lobe – it involves faith and practices. As Cacioppo explains, “The way to reach God is not through an electrode placed on the brain.”
Moreover, another study used fMRI to show that Democrats and Republicans base their support on emotions rather than rational thinking. But there are hundreds of possible ways supporters make their decisions. As Richard Henson, University College London, who wrote the paper, “What can functional neuroimaging tell the experimental psychologist?” explains, “We need more focused questions where we compare two tasks, and control for everything we can and not be confounded by other variables.”
The other problem is that cognitive functions do not exist in just one part of the brain – they can co-occur in many regions and it is the neural communication between regions that could be key. Just like harmony doesn’t exist in one area of an orchestra – it is a culmination of instruments working in intricate ways. There is no way to point a finger and say okay, the “happy” melody resides just there, next the oboe.
Certainly, the more abstract the concept the more likely it depends on this neural interconnection. Looking for the region of love in the brain might be akin to looking for the region of soul in a rhythm and blues band.
“Until recently most investigations are reporting clusters of activation in this region or that region,” says Richard Davidson, professor of psychology, University of Wisconsin. “What really may be more important is the connections of these regions, how they constitute a circuit.”
Unfortunately a timing issue comes up again when researchers attempt to study communication between regions. This high frequency connection can happen within a hundredth of a millisecond and blood flow is far too sluggish to mark it.
One trick to study the actual circuitry between networked regions is to design experiments where the communication between regions is slowed down, says Henson. Like when subjects are asked to focus attention solely on the color of a dot for a long period of time – then researchers change the hue slightly and note any change in the brain. Researchers can look at how the co-variation between two brain regions changes as a function of the color change.
Others say a progressive leap lies in squeezing out as much concrete information as possible from the fMRI signal. APS Member Frank Tong, Vanderbilt University, has been working with powerful algorithms to do just this. His team uses multivariate analysis which involves looking at patterns (or many variables) in several regions of the brain at one time, as opposed to the more common statistical method univariate analysis, which uses one variable.
Using a multivariate technique he can predict what orientation of a line a person is looking at, or even whether they are looking at a pigeon or a sparrow. But he is hesitant about calling this “true mind reading.” He’s more comfortable with the words, “brain decoding.”
“Mind reading in the true deep sense, no, we can’t do that yet,” he says. “If we ask a subject to think of anything and then predict what they are thinking about – we’re a long way off from being able to do that.”
The limiter is not in the computational methods but in the signal from the neurons, Tong says. Right now the signals are too sloppy.
“If we could measure every single neuron day in and day out, it would be shocking what we could find out,” he says. “Things like knowing when another’s mind wanders to what they ate for breakfast.”
What about knowing the integrity of another person? Some recent fMRI studies claim to show how honorable we are.
Joshua Greene gave undergraduates a moral dilemma: 1) flick a switch that will kill one person to save five people, or 2) use your hands to kill one person to save five others. The area of the brain responsible for abstract reasoning “lit up” when students face the first, less personal problem of flicking a switch. But areas associated with emotion lit up when they faced the second choice killing with their bare hands.
“This is very intuitive,” said Sabine Kastner, Princeton University, “but we can’t use this to predict that one individual will flick the switch versus knocking the person off with their bare hands.”
There are stereotypical brain patterns that can be activated by different causes, Kastner explained. The same system could light up when running into a snake on a path while hiking as when you suddenly see that the last piece of chocolate cake is gone. Snake or cake, it’s not a specific system for each emotional situation.
Greene acknowledges the industry is still in a learning curve and some of the hype has been overblown. “I think it’s really the beginning and we’re just getting our bearings,” Greene said. “What we are going to learn from imaging has yet to come – right now we are just learning our way around the brain.”
In 30 years however, he says it might be possible for a summer camp to screen their employees for say, pedophilia. The camp could put employees into a scanner, show them photos of children in bathing suits and see how their brains light up. “There are pockets in the not too distant future where we will see something that is deep, personal and socially relevant,” he said. Right now Greene is working on research about when and how people keep promises.
If this much has erupted in the last decade, where might the next decades take us?
Many scientists dream of a time when they can free their subjects from the artificial environment of the scanner. To create a sort of “hat scanner” so they can study brains at work in the real world. They also dream of seeing the finer details.
“I’m greedy,” says Randy Buckner, Harvard University. “I want to see things at the cellular level in real time – lots of them, or all of them. I want to see different aspects of cellular function and molecular function.”
Getting past the blood to see directly into a neuron is considered the greatest fantasy. But there are realistic changes underway. Standardization across the industry and strict designs for experiments might help make results more precise and free of bias. Making neuroanatomy and neuroscience requisites for a psychology degree will better train future researchers. And with the growth of computer power and the use of fusion techniques it’s quite possible that seemingly outrageous findings are within a shorter grasp than we think.
“It’s not just another fad like phrenology. It’s more than just corelational,” said Richard Henson. “With fMRI we can dynamically set up experiments and perturb them, and watch how the brain changes as function of what you are controlling experimentally.”
fMRI has been like placing the long word on a tight Scrabble board. It’s opened up the game. At once the players heave a sigh of relief and buckle down excitedly, to study all the new possibilities. And like the long word, fMRI hasn’t necessarily won the game but it might have laid the path to win.
“It’s opening the door to a whole new field,” said Kastner. “I think that if we really want to understand behavior in relation to brain function this is how we are going to do it.”