A Computational Model of the Effects of Drug Addiction on Neural Population Dynamics (1209.3411v1)

Abstract: Reward processing and derangements thereof, such as drug addiction, involve the coordinated activity of many brain areas. Prior work has identified many behavioral, molecular biological and single neuron changes throughout the mesocorticolimbic system that reflect and drive addictive behavior. Subpopulations in the ventral tegemental area (VTA) encode positive reward prediction error, negative reward prediction error, and the magnitude of the reward. Phasic activity in VTA dopaminergic neurons correlates with hedonic value. Tonic activity of groups in the dorsomedial prefrontal cortex (dmPFC) can encode antidepressant states. However, little is known about how drug addiction might affect population encoding across larger brain regions. Here, we compare the information content associated with network patterns in naive, acutely intoxicated and chronically addicted states in a plastic attractor network. We found that addiction decreases the network's ability to store and discriminate among patterns of activity. Altered dopaminergic tone flattens the energy landscape and decreases the entropy associated with each network pattern. Altered dmPFC activity produces signal-to-noise deficits similar to computational models of schizophrenia. Our results provide a conceptual framework for interpreting altered neural population dynamics in psychopathological states based on information theory. They also suggest a view of the subtypes of depression as on a continuum of combinations of cortical and subcortical dysfunction. This suggests that patients who suffer from depression with psychotic features will have more cortical than mesolimbic dysfunction. Furthermore, our framework can be applied to other psychiatric illnesses and so may help us, in general, quantitatively understand psychiatric illnesses as disorders in the representation and processing of information by distributed brain networks.