During a social “game-play” study, researchers at Baylor College of Medicine were able to figure out a person’s mental disorder based on the reactions of his or her partner. The study was conducted in an effort to find a more objective measure of mental illness.
Currently, those suffering with a mental illness such as borderline personality disorder, autism spectrum disorder, major depressive disorder or attention deficit hyperactivity disorder (ADHD) are most-often diagnosed through self-reported behavior traits.
In the study, the research team analyzed the social interaction between an ‘average’ person and a person diagnosed with a mental disorder during an ‘investment game.’
Interestingly, it was the average person’s reaction to the partner with the mental disorder that revealed the illness, said Dr. P. Read Montague, director of the Brown Human Neuroimaging Laboratory, professor of neuroscience and senior author of the report.
“The relation between social interactions and disorders is very subtle. That is why it has not been fully detected before,” said author Misha Koshelev of the Keck Center.
“In our research, sophisticated statistical algorithms running on powerful computer clusters allowed us to see disorder-related patterns behind the seemingly random social interactions. These algorithms are similar to powerful lenses that transform a blurry image into a clear picture.”
The research team observed the interaction of 287 pairs of participants who had previously participated in a simple “trust” game in which one person (the investor) is given $20. The investor could then choose to send a portion of that money to the other person (the trustee).
The amount of money sent to the trustee was tripled, and the trustee would then decide how much to send back. This continued for 10 rounds. During these interactions, the partners learned what to expect from the other person. Typically, the two never actually met or spoke.
The investor had no mental disorder but the trustee had been diagnosed with one of the following – major depressive disorder, autism spectrum disorder, borderline personality disorder or attention deficit hyperactivity disorder.
The dynamics between the two participants were classified through the following: the investment and repayment ratios, the style of play between the two participants and the dependence of the next investment on the previous ratio of investment to repayment.
“We wanted to quantify the way people interact,” said Dr. Terry M. Lohrenz, instructor in the human neuroimaging laboratory.
“We looked at 287 of these interactions and, using this data, clustered them. Then we looked to see if any of the various groups were overrepresented in the clusters, and they were.”
The clusters were formed on the investors’ reactions—not on the reactions of the participants with a mental disorder.
“They were a sort of biosensor,” said Dr. Marina Vannucci of the Keck Center and a professor of statistics at Rice University.
“We were focusing on what the investor did and his/her reaction to the other person’s response.”
After the human study, the research team developed a computer model based on the healthy investors and had it play the trust game against computerized models of the various mental disorders represented in the duos.
“We could tell a difference when the computer was playing against a computerized version of someone with borderline personality disorder,” said Lohrenz. The same applied with the other disorders.
“This opens up a whole new way of approaching diagnosis,” he said.
“Game theory has been available to mathematicians and economists for years,” said Dr. Kenneth Kishida, a postdoctoral fellow in the neuroimaging laboratory.
“Only in the past decade has it been available to neuroscientists and now we are trying to bring it into the psychiatric domain.”
Kishida and Lohrenz believe this could be a helpful tool in diagnosis but that it does not replace the proven diagnostic guidelines of psychiatry.
The study appears online in PLoS Computational Biology.
Source: Baylor College of Medicine