@jhawkins raised three unresolved problems in our theory that seem related to reference frames. He shared an idea he had just encountered about how grid cell modules overlap. If we extend that behavior into the cortex, we discussed some implications for the ability to represent unique locations, even though we have no conclusions just yet.
Summary Video
Main Video
0:00 Introduction
1:29 Three Problems That Suggest We Need to Change Our Thinking About Reference Frames
2:15 The First Problem: Shared Learning
4:30 The Second Problem: Voting Using Relative Pose
5:38 The Third Problem: Distributed Models
8:06 Question about Voting ID and Reference Frame Anchoring
12:21 One Alternative to Shared Learning
14:39 History of Grid Cell Thinking
24:02 The Moser 2016 Paper on Grid Cell Modules Overlapping
31:30 The Implications for the Cortex if Grid Cell Modules Overlap
35:40 Possible Hybrid Approach: Three Overlapping Grid Cell Modules Do Path Integration
48:57 Discussion on Overlapping Grid Cell Modules
1:02:06 @hlee Presents Primary Grid Cell Paper