This paper reports first results of a simulation of language pathology in schizophrenia. Using DISCERN, a subsymbolic model of story understanding and recall, the impact of different simulated lesions hypothesized to underlie schizophrenia is investigated. In response to excessive connection pruning, the model reproduces symptoms of delusions and disorganized language seen in schizophrenia, as well as the reduced output seen in compensated later states of the disorder. The effects of other lesions are less consistent with the symptoms of schizophrenia. The model therefore forms a promising basis for future computational investigations into the underlying causes of schizophrenia.

PDF

In Proceedings of the 29th Annual Conference of the Cognitive Science Society, 311-316, Hillsdale, NJ, 2007. Erlbaum.

@InProceedings{grasemann:cogsci07, title={A Subsymbolic Model of Language Pathology in Schizophrenia}, author={Uli Grasemann and Risto Miikkulainen and Ralph Hoffman}, booktitle={Proceedings of the 29th Annual Conference of the Cognitive Science Society}, address={Hillsdale, NJ}, publisher={Erlbaum}, pages={311-316}, url="http://nn.cs.utexas.edu/?grasemann:cogsci07", year={2007} }