Using Biophysical Computational Neural Models to Investigate Neuropsychiatric Disorders

Samuel Neymotin
The Nathan S. Kline Institute for Psychiatric Research (NKI)

Computational psychiatry has benefited greatly from the advent of neurotechnologies that allow recording of neural activity at the cellular and subcellular level. This has led to recording of enormous electrophysiological data at multiple spatial and temporal scale from multiple species under different behavioral conditions. Together with high quality anatomical data, such data has enabled scientists to build detailed models of brain circuits to dissect the mechanisms of sensory processing in healthy and diseased brain, and use these models to devise novel pharmacological and electrical stimulation treatments that transform dynamics of pathological conditions to healthy states. Over the past decade, we have built several multiscale biophysical models of Visual Cortex, Auditory Cortex, Motor Cortex and Hippocampus by incorporating anatomical and electrophysiological details to match patterns of ensemble electrical activity observed in the animals. These models have provided us many hypotheses to test e.g. we found that neuroelectric oscillatory power of the cortical models was linked with the level of information processing in the network: overly high gamma power produced stereotyped firing patterns, suggesting a loss of responsiveness to external information, and potentially explaining aspects of hallucinations in schizophrenia. Now we are extending our work by combining multiple sensory cortices (starting with Visual Cortex and Motor Cortex) interacting with visual/auditory environment in a closed-loop configuration, where the model learns about the environment by encoding visual inputs and then learns to associate rewarding behavior encoded in motor output via a reinforcement learning framework. We aim to combine these models with behavioral experiments in animals to investigate how thalamocortical circuit dynamics support audio/visual processing, and enabling predictions on the circuit-level and dynamic alterations contributing to auditory/visual hallucinations in schizophrenia patients. Our research approach may ultimately shed light on the mechanistic origin of neuropsychiatric disease and predict novel treatments for such disorders.

Presentation (PDF File)

Back to Computational Psychiatry