Fear reduction therapies (or fear conditioning) traditionally involve constant exposure to the thing that’s feared, but a new method using an artificial intelligence (AI) algorithm presents a way to rid a person of their phobia without having to expose them to it, at least not literally.
The new method was developed by researchers at the University of California, Los Angeles (UCLA), Columbia University, and the Nara Institute of Science and Technology in Japan. The team’s findings have been published in the journal Nature Human Behavior, and according to a Motherboard interview with lead researcher Hakwan Lau, “With this procedure we can condition out the fear.”
The technique uses a process called “decoded neurofeedback.” First, the team needed to identify a specific fear’s unique brain signature. Participants in the research were presented with images of colored vertical lines and given electric shocks to trigger fear when shown one particular color. Over time, that color of line became “scary” and produced a unique brain signature in the participants. These brain signatures were then given to an AI visual recognition algorithm that could abstract a pattern from them. The AI also observed further brain activity while the subject was at rest to hunt for similar patterns.
Once it had this information, the system could be used to deprogram the fear. “Whenever your brain is representing or ‘thinking about’ the red line, one of the scary things, we [tell the subject], ‘Congratulations you won 10 cents,'” Lau explained. “So, now whenever the red thing happens, instead of being paired with the electric shock, now it’s associated with a positive monetary reward.”
Every time the recognition algorithm saw even just a fleeting fragment of the fear memory, it could alert the researchers to trigger a positive reconditioning stimulus. Once the subjects were “reprogrammed,” their usual signs of fear (breaking into a sweat) were diminished when they were exposed to the once-scary line.
This research is particularly notable because specific fears generate the same brain activity in different people, with the researchers noting an 80 percent accuracy at inferring specific fears from fMRI scans (that part of their research remains unpublished). Therefore, we can create a fear library that would allow the algorithm to know fear patterns shared by many different types of people without having to actually read the brain signatures of all those people. The algorithm also doesn’t have any nefarious uses, as Lau notes that the process can’t be used to insert fear into the brain, only remove it.
Aside from being beneficial in the treatment of various fear disorders and similar ailments (like PTSD), this research also shows that the brain functions in a way that can be understood by a simple AI algorithm — one more example of technology teaching us more about the inner workings of our minds than previously thought possible.