Research at Baylor College of Medicine (BCM) shows that observing social interaction between healthy individuals and those with a mental disorder can help identify the specific disorder.
For their paper, published October 21st in the open-access journal PLoS Computational Biology, Misha Koshelev and his co-authors studied the interaction of 287 pairs of research subjects who had previously participated in a simple “trust” game, to find patterns behind the interactions.
Surprisingly, the tell-tale reactions occur in the control participant who is reacting to the partner who has a mental disorder.
Psychiatrists seeking to identify mental disorders frequently depend on imprecise measures; in this study, the authors, from BCM, the W.M. Keck Center for Interdisciplinary Bioscience Training and Rice University, aimed to find a less subjective measure of mental disorder. “The relation between social interactions and disorders is very subtle. That is why it has not been fully detected before.
In our research, sophisticated statistical algorithms…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,” said Koshelev, a Ph.D. student at BCM.
By studying the “trust” game, the goal was to quantitatively characterize the style of play exhibited by the healthy subject. To do this, the testers gave one person (the investor) $20, who could then choose to send a fraction of that amount to the other subject (the trustee). The amount sent is tripled on the way to the trustee, who then decides how much to send back.
This continues for 10 rounds. In this study, the investors had no mental disorder and the trustees had previously been diagnosed with one of several disorders – autism spectrum disorder, borderline personality disorder, major depressive disorder or attention deficit hyperactivity disorder.
Koshelev and his colleagues classified the dynamic between the two people using the numbers or amount of money exchanged. “We wanted to quantify the way people interact,” said study co-author Dr. Terry M. Lohrenz of BCM. “We looked at 287 of these interactions and, using these data, cluster them.
Then we looked to see if any of the various groups were overrepresented in the clusters, and they were.” It was found that the investor’s behavior was quantitatively and systematically influenced by their partner’s pathology.
The clustering was based on the reaction of the investor and not the person with the mental disorder. “They were a sort of biosensor,” said Dr. Marina Vannucci from Rice University. “We were focusing on what the investor did and his/her reaction to the other person’s response.”
When the researchers later computerized the behaviour of the investor and trustee, the difference when playing against someone with a borderline personality disorder remained visible.
The authors emphasized that although this provides a new way of approaching diagnosis, it does not replace the proven diagnostic precepts of psychiatry.
“We are trying to find ways to quantify mental disorders — create computational psychiatry, if you will, which will eventually help find new treatments,” said senior author Read Montague of BCM.
Funding: This work was supported by a training fellowship from the Keck Center for Interdisciplinary Bioscience Training of the Gulf Coast Consortia to MK. MV was partially supported by NIH-NHGRI grant number R01-HG003319, and by NSF-DMS grant number 1007871.
PRM was partially supported by NIH R01 grants DA11723 and MH085496. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Source: PLoS Computational Biology