Skip to main content

Scientists Find Networks Can Get Schizophrenic

The project was based on a virtual computer model designed to simulate the excessive release of dopamine in the human brain. The network was able to learn a natural language was used to investigate what happens to language as the result of eight different types of neurological dysfunction. The results of the simulations were compared by Ralph Hoffman, professor of psychiatry at the Yale School of Medicine, to what he saw when studying human schizophrenics.

The information provided to the network very much in the same way a human brain stores information-not as distinct units, but as statistical relationships of words, sentences, scripts and stories, the researchers said. “With neural networks, you basically train them by showing them examples, over and over and over again,” said Uli Grasemann, one of the project leaders. “Every time you show it an example, you say, if this is the input, then this should be your output, and if this is the input, then that should be your output. You do it again and again thousands of times, and every time it adjusts a little bit more towards doing what you want. In the end, if you do it enough, the network has learned.”

Next, the scientists reran the data input process, but simulated an excessive release of dopamine by increasing the system’s learning rate-essentially telling it to stop "forgetting" so much.  “It’s an important mechanism to be able to ignore things,” said Grasemann. “What we found is that if you crank up the learning rate in [the virtual computer model] high enough, it produces language abnormalities that suggest schizophrenia.” When retrained with the faster learning request, the network "began putting itself at the center of fantastical, delusional stories that incorporated elements from other stories it had been told to recall," the researchers said. In one example, the system "claimed responsibility for a terrorist bombing."

The scientists concluded that there are parallels between virtual neural networks and the human brain. The hop is that such computer systems can aid in clinical research in the future.