Pure Genius

Q&A: Sally Satel, psychiatrist, on the flaws of brain research

Q&A: Sally Satel, psychiatrist, on the flaws of brain research

Posting in Science

Brain imaging produces some seductive images of live activity in a person's mind. But are we interpreting them correctly?

In 2008 journalist Jeffrey Goldberg subjected himself to a series of brain scans done with functional magnetic resonance imaging (fMRI), in order to help clarify his political leaning. Turns out his brain likes President Obama very much, but it also receives pleasurable anticipation from looking at photos of Mahmoud Ahmadinejad. In the end Goldberg was left skeptical of such studies that link brain activity to opinions, behavior and emotions. "I'm not so much troubled by the question of why Ahmadinejad provoked a positive response in me -- I know what I know, despite what my brain says," he wrote in an article in The Atlantic. This of course begs the question: Can brain scans reveal some hidden truth about how we really feel?  What can they reliably add beyond what we can discern from traditional psychological and cognitive studies?

Science journals are packed with brain scan-derived conclusions about human behavior and emotion. Technologies like fMRI are truly spectacular -- we can see into the live activity of a brain while the person is engaged in tasks or looking at images or video. But many conclusions made from studies done with fMRI involve some pretty troubling leaps. Often the media can become quite seduced by the new brain research and may fuel the notion that brain imaging can uncover the truth of human nature.

Even more complicated is when such studies are brought into the courtroom or put one's personal freedom at risk. This is not to say, however, that there is not an equal number of very accurate and carefully controlled studies.

Sally Satel, a psychiatrist, and Scott Lilienfeld, professor of psychology at Emory University, have recently published Brainwashed: The Seductive Appeal of Mindless Neuroscience, which guides the reader through the benefits and limitations of new brain science. SmartPlanet spoke with Satel about what brain science can accurately reveal about the mind.

SmartPlanet: Frito-Lay commissioned a study of women's brains as they looked at their chip bags. Apparently the brain scans showed that the anterior cingulate cortex lit up, an area often associated with feelings of guilt. Researchers concluded that women felt guilty when looking at the shiny bags containing high-calorie snacks. So the company switched to matte bags in the hopes of relieving negative emotions associated with their brand. What is wrong with this strategy?

Sally Satel:  Well first there’s no guilt center in the brain. So inferring that the anterior cingulate cortex is telling you that you feel guilty is a leap.

This is one of the big issues with fMRI interpretations.

Right. One of the bigger problems with naive interpretation is something called the reverse inference problem. And what that means is -- as they did in that Frito-Lay study –- [researchers] look at a part of a brain that is differentially activated during a task, and the “task” is often having the person look at an image, listen to a sound, or be presented with a problem [to think through].

Of course you’re going to see more activity in certain areas of the brain than in other areas. But looking at the brain scan image and working backwards from that to what a person is thinking is very fraught.

Because?

For this reason: Various regions in the brain play a role in mediating many different kinds of subjective emotional states.

Could you unpack that statement a bit?

For example, the anterior cingulate is frequently cited as important in the processing of error detection or conflict. But that’s not quite the same as guilt. That’s one issue.

Another region of the brain that is frequently cited is the amygdala. It is most famous for being a fairly primitive area involved in the processing of fear. And that’s true, but it’s also relevant to the processing of novelty, surprise, anger and happiness. So, to just basically pick the emotion that suits your purpose is a problem. In the Frito-Lay case, I’m sure they imagined that women feel guilty when they eat high-fattening foods; so that was consistent with the narrative that the advertisers had imagined.

You also mention in your book that brain scans ought to tell us something we couldn’t find out from asking people directly.

Right. I’m not saying that [brain scans] do not add anything. I don’t mean that at all. But in this case it’s highly conceivable that people could express guilt associated with high-calorie food. So you have to think what does [the brain scan] add? This is very important because fMRI scanning is so expensive that if you can obtain those insights another way, it would be crazy to pay $5,000 per imaging session.

Right. The very fact that one area of the brain “lighting up” can be attributed to multiple behaviors and emotions means there has to be other variables taken into account?

Definitely. There are also other regions in the brain that are activated [during certain tasks in the lab]. But you may not be seeing them. This is a limitation that neuroscientists often acknowledge.

Yes, that could be pretty misleading.

Right. FMRI produces these striking isolated areas that are illuminated and rendered in beautiful colors, and that gives the impression that the whole brain is silent except for these few splotches.

But that is not how the brain works at all. It’s always on. Your brain is not dark except for a handful of regions. This is why superficial interpretation of brain scans has given rise to this critique that some people tend to be practicing a form of neo-phrenology, because you can’t just look at isolated bits and feel confident that you know what a person is thinking.

Researchers have been known to compare activity in the brain to a symphony, as opposed to an isolated single instrument positioned in different areas of the brain.

The brain is an enormously distributed entity. There is activity crackling back and forth all over. Various circuits are involved, and because some may be more differentially activated during tasks they’re the ones that may appear on a brain scan, but they’re not the only ones. So there’s an enormous amount of cross talk between brain areas and active circuits. Looking only at one area could be misleading.

Could fMRI be mixed with another technology to look at cross talk and circuitry?

Yes! There is one technique called decoding, or multivoxal pattern analysis, MVP. And it is very complicated.

Could you try to explain it for us?

Here’s an example: Say you look at three different objects. Maybe a picture of a car or a building. A computer learns the neural signature in your visual cortex as you observe these images. The next day you come into a lab and you look at pictures of these objects again. Now the researchers can look simply at the patterns that are captured on the scan and be able to tell with very high accuracy which image you were looking at, which is pretty incredible.

You also talk about something called neuro-redundancy.

Yes. Neuro-redundancy is the issue that the brain scan doesn’t tell you anything you don’t already know. For instance, you might see a study that says: Drug addiction in the brain. Or “envy in the brain.” Well, where did you expect it to be?

On one level, the technology is just spectacular -- that we can capture the neural activity and portray it in this way.

Right. That is why maybe so many people are surprised when they see an emotion rendered in the brain. The ability to sort of “see” the emotion is pretty amazing. A potential pitfall, however, is to observe activation in the brain and then assume that a behavior is associated with that activation, and is not under a person’s control.

Sometimes it is true that this activation pattern is defective and the behavior that flows from is not under voluntary control. But take the example of addiction. You can take someone who has a cocaine problem and put them in a brain scanner -- either fMRI or another one is PET scanning, Positron Emission Tomography -- and they both measure metabolic rates that change when a person is stimulated in some way. In this case,  researchers might present the person with pictures or videos of other people using cocaine. And they’ll experience a subjective sense of wanting it. And accompanying that is enhanced activation in the reward circuitry of the brain. That’s all a true fact, one can observe it.

But this doesn’t mean that the person is going to run out and use cocaine. The scan doesn’t tell us what the subsequent behavior will be.

This sort of thing gets dangerous in the context of the law.

It’s very important to know in a forensic setting that the behavior that flows from the observed pattern of activity in your brain is not necessarily inevitable. And I think that is sometimes obscured, and frankly in a forensic setting that’s exactly why these images are shown.

What are the best current uses for brain scans?

Brain scans are very useful in pre-surgical planning so that surgeons can [for example] remove an epileptic focus. FMRI can guide electrodes in some of the experimental treatments for deep brain stimulation in depression, OCD, even anorexia.

You can use scans to follow stroke damage. Of course, there’s Alzheimer’s. [Studies with fMRI can spot damage early so that therapies can be used to slow or maybe even prevent progression of the disease.] It would be a crime for people to think that scans are not useful.

A major point you make in Brainwashed is that problems arise when we ascribe too much importance to brain-based explanations and not enough to psychological or social ones. So how can we reconcile this?

It’s a matter of using many different kinds of brain-based technologies but certainly behavioral and psychological levels of analysis are critically important. We have to be cautious of neuro-centrism, the notion that the brain is the most important level at which to understand complex behavior.

For example, if you are interested in dopamine physiology in addiction then, yes, stay at the level of the brain. When you’re working with people in the clinic, however, focus more on behavior. So much of work with addicts is about shaping their behavior day to day so that they can avoid the cues that might incite craving.

In comparison, look at Alzheimer’s. I could not make any progress if I offered someone suffering from Alzheimer’s a deal where they could get a million dollars if their memory didn’t deteriorate. But with drug addicts we get a lot of mileage out of offering rewards of contingencies for a person changing their behavior.

Someone with the kind of brain damage you get with Alzheimer’s cannot respond to foreseeable consequences the way someone with the brain changes of addiction can. That’s a very vital difference and this difference is important to consider when considering conclusions from a brain scan.

Share this

Christie Nicholson

Contributing Writer

Christie Nicholson produces and hosts Scientific American's podcasts 60-Second Mind and 60-Second Science and is an on-air contributor for Slate, Babelgum, Scientific American, Discovery Channel and Science Channel. She has spoken at MIT/Stanford VLAB, SXSW Interactive, the National Science Foundation, the National Research Council, the Space Studies Board and Brookhaven National Laboratory. She holds degrees from the Columbia University Graduate School of Journalism and Dalhousie University in Canada. She is based in New York. Follow her on Twitter. Disclosure