Recent research published in human brain mapping provide evidence that young people experiencing suicidal ideation process death-related concepts differently in their brains compared to healthy individuals. This finding indicates that these individuals reflexively associate thoughts about death with their own feelings. The study suggests that combining brain imaging and artificial intelligence could eventually help identify people at risk of suicide based on how their brains represent certain words.
If you or someone you know is experiencing suicidal thoughts or a mental health crisis, help is available. Contact the free and confidential Suicide & Crisis Lifeline by calling or texting 988 or live chat at 988lifeline.org.
Mental health professionals typically rely on patients to report their feelings, but people at risk of suicide don’t always open up about their struggles. Finding objective physical measurements in the brain could help identify people in need of assistance.
Previous behavioral tests have shown that suicidal people tend to implicitly associate themselves with the idea of death. For example, word association games often show that these people associate words like “funeral” and “suicide” with words related to themselves. Scientists wanted to see if this psychological connection leaves a detectable footprint in the brain.
“Our lab’s goal is to understand how thinking is supported by brain activity. By examining the underlying brain activity, we investigated whether specific concepts change systematically in people who are considering suicide,” said study author Marcel Just, DO Hebb Professor at Carnegie Mellon University.
The research team is based on the idea that the human brain functions like a dictionary of universal concepts. When people think about common objects like bananas, their brains show very similar patterns of activity. The scientists aimed to determine whether the neural blueprint for death-related concepts is specifically altered in people experiencing suicidal ideation.
The study included a final sample of 154 youth between the ages of 18 and 30. Of these participants, 89 were currently experiencing suicidal ideation and 65 were healthy individuals with no history of mental health conditions. The researchers found that the two groups were equal in age, gender ratio, and general intelligence.
To measure brain activity, scientists used functional magnetic resonance imaging. This is a common brain scanning technique that tracks blood flow to different areas of the brain and reveals which areas are active during certain tasks. While inside the scanner, participants were shown a series of 28 words on a screen.
These words were classified into four categories, including suicide-related concepts, positive concepts, negative concepts, and attitude-related concepts. Words associated with suicide included death, funeral, lifeless, and hopeless words. Each word was displayed on the screen for 3 seconds, and participants were instructed to actively think about the main properties and meanings of the word.
Words were presented multiple times in different random orders. This repetition allowed the scientists to identify brain voxels, tiny three-dimensional units of brain tissue that consistently responded to specific concepts. The scientists then analyzed this brain scan data using machine learning, a type of computer algorithm designed to recognize complex patterns.
They specifically trained their algorithm to look at areas of the brain that have been associated with thinking about oneself in previous studies. These self-expression areas include structures such as the precuneus and middle temporal gyrus, which are typically activated when people think about their lives and identities.
The machine learning program was successful in differentiating between suicidal individuals and healthy participants with a moderate but reliable accuracy of approximately 57-61%. This distinction was based entirely on brain activity observed when participants thought about suicide-related concepts.
When thinking about words like “death” and “funeral,” suicidal people showed distinct activation in brain regions responsible for introspection. This pattern provides evidence that these people reflexively think about themselves when processing death-related concepts.
“People experiencing suicidal thoughts associate ‘self’ with concepts related to death,” Just told SciPost. “We can now detect these neural signatures using fMRI.”
The researchers also tested other categories of words, such as positive and negative terms. Brain activity associated with these non-suicide-related words did not differentiate the two groups beyond random chance. This specificity suggests that the altered brain patterns are strictly tied to how individuals perceive death, rather than a general difference in how they process all emotional language.
The scientists noted that the algorithm was able to distinguish between the groups even when the analysis was limited to just two words: “death” and “funeral.” They also mathematically controlled for differences in age, intelligence, and data quality to ensure that the algorithm reliably detected psychological connections to death. By identifying this specific conceptual shift, this study established a measurable neurobiological basis for suicidal ideation.
“It is technically possible to use neuroimaging to determine whether a person’s representation of death-related concepts is abnormally related to that person’s ‘self-representation,'” Just explained. “This judgment can be made subliminally, even if the person has not verbally revealed the thought.”
“Perhaps the deepest implication is that all people with healthy brains express concepts in the same way. Whether it’s ‘cup’ or ‘banana,’ the neural machinery is consistent across species. This brain’s ‘universal conceptual dictionary’ allows us to analyze activity and determine what people are thinking, a process the media often refers to as ‘mindreading.'”
Although these findings are promising, it should be noted that there are several limitations and potential misinterpretations. The accuracy rate of the current algorithm is too low for this test to be used as a standalone clinical diagnostic tool. Although this test was able to accurately identify many individuals with suicidal ideation, it also produced a significant number of false positives and false negatives.
Additionally, the scanning process requires concentration, and the researchers had to exclude data from 77 other early participants whose minds wandered during the long 25-minute task. “Performing brain activity scans in an MRI scanner is cumbersome and requires the use of very expensive equipment,” Just said. “Therefore, using this method to detect suicidal ideation is currently not practical for routine clinical screening.”
In the future, the research team hopes to improve this procedure to make it shorter and easier for participants to complete. A short task that focuses on only a few informative words could improve the quality of the data and the accuracy of the algorithm. The scientists also plan to adapt the method for use with cheaper and more accessible technologies.
For example, applying this approach to electroencephalography, a method of measuring brain waves using a sensor cap, could make the test widely available in standard clinics. Ultimately, developing treatments that help break the psychological link between self and death could provide new avenues to support people at risk of suicide.
The study, “Neural representations of death-related concepts identify conceptual changes in the self in suicidal youth,” was authored by Marcel Adam Just, Robert Mason, Lisa Pang, Dana McMakin, Kristin Cha, Matthew K. Nock, and David Brent.
