Researchers at U. of Texas and Yale Use Computers to Simulate Schizophrenia
From: The Chronicle of Higher Education | Researchers at U. of Texas and Yale Use Computers to Simulate Schizophrenia
Computer simulations of malfunctioning brains may be the key to understanding schizophrenia and other conditions.
A research team including computer scientists at the University of Texas at Austin and a professor of psychiatry at Yale have been testing various theories of how schizophrenic brains misfire as they process information. People with schizophrenia often have trouble repeating different stories, for instance, frequently combining elements of separate stories and inserting themselves into the narrative.
“We are trying to quantify being delusional,” says Risto Miikkulainen, a professor of computer science and neuroscience at Texas, who designed the computer network on which the hypotheses were tested.
The results of their work will be published in the May 15 edition of Biological Psychology.
The team’s simulation suggested that the problem might be a phenomenon called “hyperlearning”—meaning that people with schizophrenia focus too much attention on recent pieces of information and fail to successfully integrate this information with prior learning. This results in an inability to properly identify what information is most important.
“Hyperlearning means that memory consolidation is abnormal,” says Mr. Miikkulainen. “It doesn’t form a coherent structure of the world.”
For the computer network, learning how to build words into sentences, subplots, and stories is a gradual process that researchers develop by feeding the network example after example. This process teaches the computer what information is most important and how to construct stories out of these building blocks.
The networks following the hyperlearning model were trained normally until the very end, when the “rate” of learning was increased, which threw off the balance of the earlier training.
When given new stories to build and repeat, the hyperlearning networks recalled fewer of the stories than other networks, often mashed together elements of different stories, substituted the wrong words, and frequently inserted the wrong characters into stories.
Inserting the wrong characters in stories explains why people with schizophrenia sometimes perceive themselves to be persecuted or have delusions of grandeur—they insert themselves into other people’s stories.
Mr. Miikkulainen says the team, which included Uli Grasemann, a Texas graduate student, and Ralph E. Hoffman, a professor of psychiatry at Yale’s medical school, wasn’t expecting the model to identify hyperlearning as the most plausible hypothesis and, rather, anticipated that schizophrenia was more likely caused by an excessive “pruning,” or destruction, of neural networks as people age.
He says the next step is to test the findings of the model more on human patients using brain scans and interviews, but he also hopes to use his program to study other mental illnesses—such as depression and bipolar disorder—and to use the model to test various treatments for schizophrenia and these other illnesses.
- Awards & Honors
- About Us
- Student Engagement and Support
- Masters Program
- Ph.D. Program
- Financial Information
- Prospective Students
- Incoming Students
- Current Students
- Portfolio Program in Robotics
- Curricular Practical Training
- Grad Student Talks
- UTCS Direct