Room Location: Salomon 001

Held at the 5th Multidisciplinary Conference on Reinforcement Learning and Decision Making (RLDM 2022)

Saturday, June 11th, 2022 - Brown University, Providence, Rhode Island, USA

Replaying the past and imagining the future can accelerate learning and facilitate flexible behavior. However, mental simulation can also be dysregulated, which in humans may lead to debilitating forms of repetitive thinking, such as chronic worry and rumination, and in machines can result in biased value functions and suboptimal policies. This workshop will bring together computer scientists, computational neuroscientists, and clinical scientists to address key open questions and challenges concerning what distinguishes helpful from harmful mental simulation. Topics will include how simulation may be biologically realized (e.g., the neural bases of replay), how agents learn to initiate and terminate simulation, how sampling experience can accelerate learning, and how sampling state representations can facilitate abstraction. Throughout, we will consider how these topics can shed light on pathological repetitive thinking processes in humans — and, in turn, how understanding such naturalistic processes can offer insights into the mechanisms, constraints, and challenges that both minds and machines face when they engage in simulation. A key aim of this workshop will be to establish a network of interdisciplinary researchers. Our long-term goal is to develop a general computational framework that can both explain the role that simulation plays in basic learning mechanisms and account for how it becomes dysregulated in various forms of repetitive negative thinking in humans, with implications for numerous psychiatric disorders.