Unarmed Black civilians are three times more likely to be shot and killed by police officers than unarmed white civilians in the U.S. In tragic cases in recent years, unarmed Black men holding innocuous objects like a wallet, cell phone, or vape pen were killed by police officers because those objects were misidentified as weapons. These split-second fatal mistakes, often under ambiguous and stressful conditions, have sparked urgent debates about their causes and how to fix them.
A new brain-imaging study from researchers at Columbia University suggests that part of the problem is that racial stereotypes can infiltrate the brain's visual system, prompting us to see objects in ways that conform to these stereotypes. These stereotypes transiently distort how the brain quite literally sees a harmless object. Using functional magnetic resonance imaging (fMRI) and cutting-edge neural decoding techniques, the researchers found that when participants saw everyday graspable objects like a wrench or drill after briefly viewing a Black man's face, object-processing regions in the brain shifted their neural representation to more closely resemble that of a weapon.
Participants were also asked to sort images, identifying them as either weapons or tools. They consistently showed milliseconds of delay in categorizing tools as tools rather than as weapons when the image was immediately followed by a Black man's face, indicating an initial unconscious tendency to perceive them as weapons. The researchers then specifically linked this weapon identification bias to the shifts in the brain's visual system they observed: The more that test subjects' brains shifted toward a weapon reaction when they saw a tool followed by a Black man's face, the longer the delay they experienced before successfully resolving the tool as a tool, not a weapon. In some cases, particularly when responding very quickly, subjects made full-blown errors such that tool images followed by a Black man's face were misidentified as weapons altogether.
The study was led by Jon Freeman, an associate professor of psychology at Columbia, and published in the journal Nature Communications. The other authors, previously in Freeman's lab group at Columbia, include DongWon Oh, an assistant professor of psychology at National University of Singapore , and Henna Vartiainen, a doctoral candidate at Princeton University .
"Our findings demonstrate that the stereotypes we hold can alter the brain's visual representation of an object, distorting what we see to fit our biased expectations," Freeman said.
The findings were robust, replicated across two large and racially diverse samples, and occurred in participants of all racial backgrounds, consistent with the fact that these stereotypes are learned by all members of society. The study included fMRI imaging of 31 subjects from the New York City area, and tested racially biased weapon identification on 422 additional online subjects based around the U.S. The researchers compared study subjects' reactions to images of Black and white men.
These kinds of racial bias effects on weapon identification have previously been observed in samples of both civilians and police officers, but this is the first brain-imaging study to look into the mechanisms involved. Previous research has linked weapon identification biases with stereotypes that associate Black individuals with crime and danger. Researchers have largely believed that these weapon identification biases do not stem from any visual distortion, instead assuming that people see the object accurately but are then unable to control their racially biased impulses. This new study suggests that an overlooked part of the problem is that stereotypes are creating a temporary visual distortion as well.
By showing that stereotypes can alter perception itself—not just conscious decisions—the findings add a crucial layer to understanding why racially biased mistakes happen in high-stakes contexts like policing.
The work opens the door to new interventions that could target visual perception, which the researchers plan to explore in future research. For instance, repeatedly pairing images of Black men with everyday tools rather than weapons could retrain the visual system, weakening the automatic bias to see a weapon where none exists. Counterintuitively, the opposite may also be true: much like how the eyes adjust after staring into bright light, prolonged exposure to Black-weapon pairings might fatigue the bias itself, allowing neutral objects to be perceived more accurately. The team is planning to explore which possible remedies yield the greatest success.
"We're eager to build on this research by exploring new interventions that might recalibrate biased visual perceptions," Freeman said, noting that traditional bias-reduction strategies have fallen short. "Our findings suggest a new direction: targeting not just the stereotypes people hold, but also the visual processes that shape how we see others. If we can change split-second perceptual distortions, we may be able to mitigate these kinds of consequential misjudgments in high-stakes situations under stress and uncertainty."
