About a decade and a half ago, the neuroscience world got super-stoked about a sexy new way to look at living brains: functional magnetic resonance imaging. It watches oxygen in blood flow to different parts of the brain and uses that as a proxy for activity---and generates lovely images of brains at work.
In the years since, fMRI has given neuroscience a bit of a bad reputation, thanks to some shoddy statistical work and an unfortunate study in which Dartmouth scientists put a dead salmon in an MR tube ... and saw neural activation. The field has largely moved past the follies of its youth. But the allure of fMRI still seems to be too much for some scientists to ignore, despite its limitations. A study out of Stanford today---a collaboration between its design school and its brain sciences research center---used the method to search for brain activity associated with visual creativity. Their test? Sticking subjects into a magnetic resonance tube while they played Pictionary.
Now, fMRI is still a great tool---just as long as you’re applying it to questions that it can actually answer. The problem is, a lot of the questions that can be answered simply with fMRI data have, by virtue of being simple, already been answered. That means any successive studies done with fMRI have to meet a much, much higher bar.
This, unfortunately, is not one of those studies.
Like many experiments conducted in the early days of fMRI research, the Stanford work, published today in Nature's open access journal Scientific Reports, uses an “A vs. B” design. It takes two conditions---in this case, one in which study participants hear a word and then draw a picture inspired by it on an MR-safe tablet (cool!), and one in which they draw a control figure, a zigzag. Then it compares the brain activation produced by each of those activities. If Pictionary is an example of visual creativity, then the parts of the brain that contribute to creativity are the ones that are activated when you overlap all those brain images and subtract the activation seen during the control task, the zigzag.
Now, for this to work the control task has to be very, very carefully designed. It should be identical to the test condition in every way---except for the function you’re interested in, which in this case is creativity. “The main problem here is that the control task wasn’t a great control task,” says Lisa Aziz-Zadeh, a cognitive neuroscientist who studies creativity at the University of Southern California. A zigzag is a lot easier to draw than many of the other items the study asked participants to draw. So the brain scans could have been showing the results of a task that’s simply more difficult to carry out. “It’s also not matched as far as language,” says Aziz-Zadeh. “Pictionary is very language-driven task, and zig-zag is not.”
A well-matched control would have been especially important in this case, because the Stanford study purports to connect an entirely new and unexpected part of the brain to creative processing: the cerebellum, which neuroscientists traditionally think of as the center for motion. Any study that tries to nail the cerebellum to creative actions has to work really hard to exclude any motion effects from the task. By using a drawing task, “it’s obvious that that the cerebellum is going to get involved,” says Adam Green, a cognitive neuroscientist at Georgetown University. “In order to pull this off they’d have to have the ideal control task.”
To be fair, the Stanford researchers recognized the difficulty in choosing a great control---especially after one of their anonymous reviewers pointed out the zigzag’s flaw. “That was a hard problem for us,” said Manish Saggar, the study’s first author. “We considered drawing calligraphy, or connecting the dots.” But both Green and Aziz-Zadeh think a better control alone wouldn't have solved the study’s deeper problems.
Like what? Well, interdisciplinary collaborations are often a good thing, especially between science and the arts. It makes sense that a design scholar would want to know how creativity works---in this case, a person who teaches a creativity course at Stanford's d-school actually suggested the study. But creativity is, in the end, a human construct. That lack of definition makes it tough to study, even though the researchers tried to focus on a specific kind.
Now, none of this is to say that implicating a new brain area in creative processing is a bad thing. Challenging pre-established notions is kind of what science is supposed to do. And in fact, a lot of scientists are really excited about the possibility of redefining the cerebellum’s long-held motor role. “The cerebellum shows up a lot of time when you’re doing neuroimaging research,” says Green, and it’s often difficult to account for, since most fMRI tasks are designed to explicitly limit movement---any movement can affect the clarity of the signal picked up by the machine.
Instead of purely controlling movement, then, maybe the cerebellum is more of a coordination system. That's a theory proposed by neuroscientist Masao Ito. “To apply what the cerebellum does for the motor system to cognitive processes that are complex---that have multiple moving parts, that need to be choreographed---makes a lot of sense,” says Green. And indeed that's what Saggar and his colleagues are interested in exploring.
But that’s a massively complex question---as many of the remaining questions in cognitive neuroscience are. The larger problem here is that it seems fMRI’s moment of utility, at least in the sense of finding particular neural correlates of a high-level process like creativity, has passed. The method is really good at looking at two conditions and finding the differences between them. “But if the answer was simple, it would have already been done by now,” says Ayse Saygin, a cognitive neuroscientist at UC San Diego. “If there was a creativity area that was reliably activated it would have been found.”
For a long while, neuroscience has been dominated by theories of domain specificity---linking a specific part of the brain, like Broca’s area in the frontal lobe, to a particular function, like speech. That’s because functional characterization was all that early neuroscience could do. In the 19th century, researchers had to wait for a patient with a specific defect, like aphasia, wait for them to die, and then look at their brain. Whatever area of the brain was damaged was the one that contributed to that specific skill.
This research became massively simpler when fMRI started allowing researchers to study brain activation during specific activities in living patients. But it also helped them to realize that domain specificity wasn’t all it was cracked up to be. Most researchers have accepted that the brain is a lot more complicated than “x does y.” “If you took some article from between 2000 and 2005 versus the last 5 years and compared how much they used region and area-focused vs. network, there’s so much more network now,” says Saygin.
What that means is that neuroscience is waiting for new methods. “Now all of the A vs. B studies are done,” says Saygin, “and in parallel, computation has advanced.” Instead of fMRI, which provides static images---not changes in brain activation over time---researchers like Saygin are using computer modeling to understand the causal connections between different parts of the brain.
And new methods can start to understand circuits with more specificity by directly activating them---with so-called neurointervention studies. Green uses transcranial direct current stimulation, zapping parts of the brain through the skull and observing the resulting changes in behavior. “It’s one thing to see an area light up,” says Green. “But now you can experimentally manipulate an area and see an effect.” Maybe one of those new methods would have helped Saggar and his colleagues design a study that really told us what goes on in Pictionary-playing brains. Or maybe there are just some questions that we can’t answer yet.
