Does the Neocortex Use Grid Cell-Like Mechanisms to Learn the Structure of Objects?
summary
A concrete claim: the neocortex is a location processor where every cortical column uses grid cell-like location signals (proposed in layer 6A) to learn objects as sets of features at locations. Numenta builds simulation models with neurons that have active dendrites, sparse codes (e.g. 5,000 neurons with ~2% active gives astronomic capacity), and unsupervised local learning; these sequence memory and sensory‑motor inference models learn hundreds of thousands of transitions, predict sensory inputs during movement, support compositional objects (6A/6B + layer 5), and are validated in simulation and published work.
AI-written summary · about
chapters (23)
- 0:00Introduction
- 7:32Understanding the Neocortex
- 13:14Outline
- 16:50Learning Sequences
- 19:03Sparse Activations
- 21:52Single Layer Model
- 23:42The Problem
- 26:08The Object Layer
- 27:04Maps
- 28:28Multiple Columns
- 30:09Location in Grid Cells
- 32:28Location in Objects
- 35:16Timescale
- 37:11Hierarchy
- 39:31Unions
- 40:21Hierarchy Challenges
- 43:11Border Ownership Cells
- 45:26Location Processor
- 47:06Core Research Team
- 48:36Mathematically expedient
- 49:46What and where pathways
- 51:46Local learning rules
- 52:56Color
related talks
30mCracking the code: neuronal networks in the brain
57mWhat Should a Computational Theory of Cortex Explain?
51mComputing like the Brain: the Path to Machine Intelligence
51mCognition and Computation
27mWhat can artificial neural networks learn from biological neuromodulatory systems?
58mStrong Generalization from Small Brains and No Training Data
13mEnhanced EEG Data Classification with 3D SNN and On-Line Learning ESN
59mIntelligence and Machines: Creating Intelligent Machines By Modeling the Brain