@nleadholm recaps the brainstorming week on actions and causality, exploring how the brain could learn to flip a switch using behavior and morphology models. The team discusses forward and inverse causality, goal states, and how linked behaviors guide intelligent action in complex environments.
Summary Video
Main Video
0:00 Introduction
1:12 Summary of the Brainstorming Focus Week
3:29 Establishing Required Connectivity
9:44 How Could Two Correlated Behaviors be Separated
13:52 Recap of Behavior and Morphology Models in a Hierarchy
15:51 Changes in Child Object as Inputs to Behavior Models
21:36 How Could Object Movement be Deduced in Touch?
23:22 Child Object Orientation Changes also go Through the Thalamus
27:23 Flipping Up and Flipping Down are Two Behaviors
34:39 Having the Child Object Maybe Useful for Deriving Direct Motor Projection
39:15 Converting Motor Output into Egocentric Reference Frame
1:05:59 The Role of Feedback Connections for Achieving Goals
1:18:37 The Two Paths
1:21:36 Path 1: Learning Causal Links Between Models
1:26:12 Path 2: Motor Output to Subcortical Regions
1:30:35 A Real World Example
1:32:08 Actions that Require Attention and Planning vs. Practiced Movements