Brain scans reveal the vocabulary that drives neural polarization
University of California - Berkeley
Researchers
from the University of California, Berkeley, Stanford University and Johns
Hopkins University scanned the brains of more than three dozen politically
left- and right-leaning adults as they viewed short videos involving hot-button
immigration policies, such as the building of the U.S.-Mexico border wall, and
the granting of protections for undocumented immigrants under the federal
Deferred Action for Childhood Arrivals (DACA) program.
Their
findings, published today in the Proceedings of the National Academy of
Sciences, show that liberals and conservatives respond differently to the
same videos, especially when the content being viewed contains vocabulary that
frequently pops up in political campaign messaging.
"Our study suggests that there is a neural basis to partisan biases, and some language especially drives polarization," said study lead author Yuan Chang Leong, a postdoctoral scholar in cognitive neuroscience at UC Berkeley.
"In particular, the greatest differences in neural activity across
ideology occurred when people heard messages that highlight threat, morality
and emotions."
Overall, the results offer a never-before-seen glimpse into the partisan brain in the weeks leading up to what is arguably the most consequential U.S. presidential election in modern history.
They underscore that multiple factors, including personal experiences and the news media, contribute to what the researchers call "neural polarization."
"Even when presented with the same exact content, people can respond very differently, which can contribute to continued division," said study senior author Jamil Zaki, a professor of psychology at Stanford University.
"Critically, these differences do not imply that people are hardwired to
disagree. Our experiences, and the media we consume, likely contribute to
neural polarization."
Specifically,
the study traces the source of neural polarization to a higher-order brain
region known as the dorsomedial prefrontal cortex, which is believed to track
and make sense of narratives, among other functions.
Another
key finding is that the closer the brain activity of a study participant
resembles that of the "average liberal" or the "average
conservative," as modeled in the study, the more likely it is that the
participant, after watching the videos, will adopt that particular group's
position.
"This
finding suggests that the more participants adopt the conservative
interpretation of a video, the more likely they are to be persuaded to take the
conservative position, and vice versa," Leong said.
Leong
and fellow researchers launched the study with a couple of theories about how
people with different ideological biases would differ in the way they process
political information. They hypothesized that if sensory information, like
sounds and visual imagery, drove polarization, they would observe differences
in brain activity in the visual and auditory cortices.
However,
if the narrative storytelling aspects of the political information people
absorbed in the videos drove them apart ideologically, the researchers expected
to see those disparities also revealed in higher-order brain regions, such as
the prefrontal cortex. And that theory panned out.
To
establish that attitudes toward hardline immigration policies predicted both
conservative and liberal biases, the researchers first tested questions out on
300 people recruited via the Amazon Mechanical Turk online marketplace who
identified, to varying degrees, as liberal, moderate or conservative.
They
then recruited 38 young and middle-aged men and women with similar
socio-economic backgrounds and education levels who had rated their opposition
or support for controversial immigration policies, such as those that led to
the U.S.-Mexico border wall, DACA protections for undocumented immigrants, the
ban on refugees from majority-Muslim countries coming to the U.S. and the
cutting of federal funding to sanctuary cities.
Researchers
scanned the study participants' brains via functional Magnetic Resonance
Imaging (fMRI) as they viewed two dozen brief videos representing liberal and
conservative positions on the various immigration policies. The videos included
news clips, campaign ads and snippets of speeches by prominent politicians.
After
each video, the participants rated on a scale of one to five how much they
agreed with the general message of the video, the credibility of the
information presented and the extent to which the video made them likely to
change their position and to support the policy in question.
To
calculate group brain responses to the videos, the researchers used a measure
known as inter-subject correlation, which can be used to measure how similarly
two brains respond to the same message.
Their
results showed a high shared response across the group in the auditory and
visual cortices, regardless of the participants' political attitudes. However,
neural responses diverged along partisan lines in the dorsomedial prefrontal
cortex, where semantic information, or word meanings, are processed.
Next,
the researchers drilled down further to learn what specific words were driving
neural polarization. To do this, they edited the videos into 87 shorter
segments and placed the words in the segments into one of 50 categories. Those
categories included words related to morality, emotions, threat and religion.
The
researchers found that the use of words related to risk and threat, and to
morality and emotions, led to greater polarization in the study participants'
neural responses.
An
example of a risk-related statement was, "I think it's very dangerous,
because what we want is cooperation amongst the cities and the federal
government to ensure that we have safety in our communities, and to ensure that
our citizens are protected."
Meanwhile,
an example of a moral-emotional statement was, "What are the fundamental
ethical principles that are the basis of our society? Do no harm, and be
compassionate, and this federal policy violates both of these principles."
Overall,
the research study's results suggest that political messages that use
threat-related and moral-emotional language drive partisans to interpret the
same message in opposite ways, contributing to increasing polarization, Leong
said.
Going
forward, Leong hopes to use neuroimaging to build more precise models of how
political content is interpreted and to inform interventions aimed at narrowing
the divide between conservatives and liberals.
In
addition to Leong and Zaki, co-authors of the study are Robb Willer at Stanford
University and Janice Chen at Johns Hopkins University.