Full Program »
Neurocomputational approaches to depression and borderline personality disorder
This symposium will review novel neurocomputational approaches to understanding maladaptive behavior and emotional instability in depression and borderline personality disorder (BPD). The broad goal of the symposium is to provide participants an overview of reinforcement learning (RL) models and experimental paradigms that can help to understand decision processes in mood and personality disorders.
Vanessa Brown (Pitt) will provide an introduction to hierarchical Bayesian approaches to estimating decision parameters in RL models. She will articulate the value of such methods to understanding individual differences and psychopathology. Alex Dombrovski (Pitt) will present a new RL model of learning in a large, complex environment. In model-based fMRI analyses using voxelwise deconvolution, he will describe how posterior hippocampus encodes local reinforcement history and promotes exploration, whereas anterior hippocampus encodes the global value maximum for the entire environment and supports convergence on it. Building on this approach, Michael Hallquist (Penn State) will speak about the effects of emotional cues on decision-making and information dynamics in BPD. His talk will demonstrate how in BPD, greater sensitivity to recent rewards in posterior hippocampus disrupts the development of a global value representation. Finally, Michael Treadway (Emory) will present a multi-modal imaging study (MRS and fMRI) examining the interaction of acute and chronic stress on mPFC glutamate and value encoding during an RL task in depressed patients.
Talks will include non-mathematical summaries of models as well as key equations for interested members of the audience. We will provide a few minutes for questions after each talk and will conclude the symposium with 15-20 minutes of general discussion.