Detailed modeling in the Computational Core will address hypotheses concerning the specific projects in this grant, linking the cellular, systems, and behavioral levels. Our central analytical tool will be the hippocampus-entorhinal cortex model that was developed in the last funding period. We will increase its degree of biological realism in order to accommodate new research findings generated by the individual projects. Its purpose is to shed light on the hippocampal etiology of schizophrenia at an intermediate or """"""""endophenotype"""""""" level. Because it is important that our model manifest behaviors that are clinically verifiable and that capture important symptomatic features of schizophrenia, a number of psychological tests will be used as assays of schizophrenia vs. normal functioning, including The California Verbal Learning Task (CVLT), modified free recall tasks with semantic cueing, the paired associate (PA) task, and the crossing sequences (CS) task. The computational work related to this grant we have done to date has indicated that a failure to appropriately utilize context is a core cognitive abnormality in schizophrenia;the category clustering indices that are part of the CVLT tap this cognitive function. The PA task is thought to index declarative memory function generally, and schizophrenic patients have consistently shown difficulties with it. It is thought that inability to recall and process sequential information, as revealed by deficienies in the CS task, may underlay the tangentiality and looseness characteristic of schizophrenic thought. After training the network on these tasks, we will subject it to the the many neurophysiologic or neuroanatomic abormalities thought to cause or contribute to schizophrenia (e.g., interneuron deficiencies and related connectivity disturbances, decreased NMDA activity as a result of abnormal levels of NAAG, excessive dopamine levels). We predict that schizophrenogenic changes will produce decreased performance of the model on the aforementioned psychological tasks. We then will examine model behaviors to test hypotheses, as described in the individual projects of the grant, concerning the mechanisms by which the cellular level abnormalities interact to produce clinical symptoms. Even if particular hypothesized mechanisms do not appear to be operative, this is instructive also--as biologically realistic in silico modeling is completely transparent, its behaviors can generate new hypotheses which can then be tested experimentally.
Showing the most recent 10 out of 128 publications
