BI ViDA Panel:
Deep RL and Dopamine
Before the panel, Matt Botvinick gave a talk about how deep reinforcement learning is doing amazing things and helping drive our understanding of brain function.
Different views on dopamine, deep RL, and the brain
What can artificial intelligence teach us about how the brain uses dopamine to learn?
Recent advances in artificial intelligence have yielded novel algorithms for reinforcement learning (RL), which leverage the power of deep learning together with reward prediction error signals in order to achieve unprecedented performance in complex tasks. In the brain, reward prediction error signals are thought to be signaled by midbrain dopamine neurons and support learning. Can these new advances in Deep RL help us understand the role that dopamine plays in learning? In this panel experts in both theoretical and experimental dopamine research will discuss this question.
0:00 – Intro
4:45 – Armin Lak
6:50 – Ashok Litwin-Kumar
8:28 – Ilana Witten
10:15 – Ida Mommenejad
11:57 – Is reward enough?
14:56 – Can AI lead neuroscience astray?
25:45 – Other roles for dopamine
31:33 – Does structure matter?
44:30 – Non-brain-like solutions in AI
48:03 – What’s missing in deep RL for brains?