This is Your Brain on Brain Scans

Neuroscience is now drawn into many topics, including substance abuse, politics, education, economic decisions and antisocial behavior.

Brain science started getting popular with PET (Positron Emission Topography) scans in the 1990s, and then even more widespread with fMRIs (functional Magnetic Resonance Imaging) throughout the 2000s.

The scans with colorful, bright blobs displaying “activity” seem to paint a picture of the brain’s response to various stimuli, but the reality is more complicated.

We’re going to dive into how these brain scans work and the different ways that they are misleading.

How Brain Scans Work

Most brain scans popularized in the media use fMRI, which shows active regions of the brain by tracking changes in the flow of oxygen-rich blood. The underlying assumption is that the areas receiving an increased flow of blood during a cognitive task mean they are more active. Researchers then examine correlations between the regions of the brain that “light up” during these tasks.brain scan image

Brain Scans can be Misleading

Is it really as simple as scanning the brain during a task to figure out what parts of the brain are responsible for those abilities? Here are eight perspectives that explain why brain scans can be problematic:

Multiple Comparisons Correction

“I need a whole Atlantic salmon. For science.” Those were the words of Craig Bennett, a neuroscientist who had a little fun and on the way pointed out a central issue with fMRI scans. In 2009, Bennett and his fellow Dartmouth researchers placed a salmon in the fMRI and ran what should have been a silly experiment. They asked the salmon to respond to images of people in social situations by saying what the person in the situation must feel.

In fMRI studies, the information of the brain is broken down into sections called voxels, and these scientists look at which voxels are activated to create comparisons. When thousands of comparisons are created, as they are in these studies, a statistical inference issue called the “multiple comparisons problem” happens.

Basically, errors in inference, including false positives, are more likely to occur when a set is considered as the whole. As the number of comparisons increases, so can the number of perceived differences.

In a typical fMRI scan 130,000 voxels are compared, which creates a high probability of false positives, as Bennett’s poster states. While correction for multiple comparisons can help rule out these false positives, this measure is often ignored by investigators. He proved just how important correction for multiple comparisons is statistically because by not doing so, he found a cluster of activity in the dead salmon’s brain.

Obviously, the dead fish could not react to the images, let alone provide a response to human emotions. Once a statistical correction for false positives was utilized, no active voxels were found.

‘Neuro’ Sounds Legitimate

When studies say “brain scans indicate,” people listen. Or at least that’s what a Yale study found when presenting brief descriptions of psychological phenomena followed by different explanations.

The researchers created good and bad explanations, and then presented these explanations with or without irrelevant neuroscience information to regular adults, neuroscience experts and students in a neuroscience course. Although experts confirmed that the neuroscience was irrelevant to the explanation, normal adults and students in a neuroscience course found the explanations with the irrelevant neuroscience information were more satisfying than the explanations without.

Including neuroscience babble especially impacted normal adults’ judgments of bad explanations, masking otherwise perceived problems in them.

Media on the Brain

It doesn’t help that the media tends to distort, generalize and embellish neuroscience findings. In one study, researchers examined the headlines, leading paragraphs and body text terms of nearly 3,000 neuroscientific articles published from 1995 to 2004 and found that even in articles identifying as research reports, there were not many details about the actual research.

Overall, the researchers found that logically irrelevant neuroscience provides arguments with unfounded scientific credibility.

Another study found three ways the media wrongfully interprets neuroscience:

  • Neuro-essentialism, which depicts the brain as the essence of a person, and synonymous with the concepts of person, soul or self
  • Neuro-realism, which makes subjective experiences like love, pain and addition into an objective phenomenon by offering visual proof through “neuroimages” of the experience
  • Neuro-policy, which uses neuroscience to support political agendas

Mind Your Methods

widely-cited critique of fMRI from 2009 that appeared in Perspectives on Psychological Science questioned statistical methods used in brain scans. The authors looked at 55 social cognitive neuroscience studies and determined that the studies provided impossibly high correlations between brain activity and personality measures.

While most of the correlations were deemed “puzzling,” because the methods sections of the studies examined did not include how the correlations were obtained, more than half did acknowledge using a strategy of examining correlations for individual voxels that can produce misleading, inflated correlations.

The Impact of Statistical Power

2013 study examined neuroscience meta-analyses published in 2011 and found that the average statistic power of studies in neurosciences is very low. Issues with statistical power include:

  • A reduced chance of detecting a true effect
  • A reduced likelihood that a statistically significant result reflects a true effect
  • An overestimate of effect size
  • Low reproducibility of results

The study deemed research with low statistical power as “inefficient and wasteful.”

Correlation Does Not Equal Causation

This concept is a much-emphasized one for any psychology student, but it still needs noting in the case of brain scans. Even if there is a correlation between a particular cognitive task and a part of the brain, the idea that when the brain pattern is observed that task must be happening takes the correlation too far.

For example, University of Michigan researchers found that the brain can’t differentiate between emotional and physical pain, as the same areas of the brain light up whether you’re thinking of an undesirable breakup or being burned by hot coffee.

Cognitive Function is a Team Sport

Different areas of the brain are activated for various cognitive functions. Reducing emotion to only the amygdala or decision-making to only the prefrontal cortex is an abbreviated, and potentially misleading statement.

For example, comprehending a sentence probably involves Broca’s area, the spot often associated with language, but it also may include part of the sensorimotor cortex, the basal ganglia and the temporal lobe. How this complex network of neural tissue works together is what matters.

Voxels Aren’t That Small

Voxels are the smallest element of a brain image that an fMRI can see, but voxels are bigger than neurons. Studying how neurons cooperate within those voxels could be a more useful approach, since most of the action in the brain is at the neuron level.

The Brain versus The Mind

Clearly from media representation to statistical issues, brain scans can be a misleading oversimplification of some of our highest concepts. Although it’s tempting to look at simple solutions for issues like addiction and crime, these are much more difficult to examine than just looking at which area of the brain blood flows to.

Neuroscience reductionism often fails to consider the complexities of the human mind. Examining behavior and mental processes in conjunction with brain scans can help create a holistic view of cognitive functions, but brain scans alone should not be used to exemplify complicated mental tasks.

Explore Programs