UT Researchers Afflict Computer With Schizophrenia

BY Jordan Schraeder in 40 Acres on June 24, 2011
UT Researchers Afflict Computer With Schizophrenia

Using a computer to model an illness of the human mind may seem counterintuitive. Yet a pair of UT researchers say they can gain insight into schizophrenia by doing just that.

Postdoctoral researcher Uli Grasemann and his advisor, Risto Miikulainen, used DISCERN, a computerized artificial neural network, to mimic the effects of excess dopamine on the brain. They hypothesized that dopamine may affect the brain’s ability to identify whether information is or is not relevant

DISCERN, created by Miikulainen, has the ability to remember and recall simple stories and grammar. To simulate an excessive dopamine release, the pair increased the computer’s learning rate, making it difficult for DISCERN to “forget” details.

The result: the computer began to tell wild stories, even claiming it was responsible for a terrorist bombing. In another instance, DISCERN began showing signs of a condition called derailment, using jumbled sentences and constantly flipping between the first- and third-person perspective.

Both of these manifestations support the hyperlearning hypothesis, which speculates that schizophrenic symptoms are caused by the inability to choose between important and unimportant information.

And according to scientists, this representation of the disorder accurately reflects the symptoms of actual patients. People with schizophrenia often fall victim to delusional false connections between bits of information, just like DISCERN did when it claimed to be a terrorist.

The implications of this research are astounding. With additional clinical trials, this new evidence may bring scientists closer to nailing down which parts of the brain are affected by schizophrenia and other mental disorders.

Tags: