Levels of understanding

There are various perspectives on the levels of understanding, each offering unique insights:

1. Fundamental Equations and Emergent Patterns:

   • Sean Carroll discusses how understanding can range from knowing the fundamental equations governing reality to recognizing emergent patterns. For example, having a neuron-level understanding of the brain is different from understanding high-level behaviors and consequences 1 .

2. Continuum of Understanding:

   • Thomas Dietterich and Daniel Bashir describe understanding as existing along a continuum, illustrating how deep learning's interpretability challenges highlight varied depths of comprehension 2 .

3. Machine Learning and Systematic Understanding:

   • Tomaso Poggio emphasizes that even if we can construct learning machines, we might not fully grasp what they discover. He illustrates this with the concept of understanding ethics as both a human and computational challenge 3 .
     
     1
     Levels of Understanding
     01:18:56
     
     2
     Understanding Continuum
     00:20:11
     
     3
     Understanding Learning Machines
     01:01:20
     
     4
     Layers of Understanding
     00:03:52
     
     5
     Levels of Analysis
     01:00:44
     
     6
     Levels of Compassion
     00:42:04
     
     7
     Levels of Information Processing
     00:42:58

4. Categorization and Perception:

   • Jordan Peterson talks about understanding as mapping various realms (material, behavioral, imaginative, and linguistic) and how they interact holistically to form a complete understanding 4 .
   • He also mentions analyzing phenomena at various hierarchical levels to determine what analysis provides maximal functional grip, relating this to practical utility in understanding 5 .

5. Compassion and Subjective Connection:

   • Gabor Maté describes multiple levels of compassion, from understanding others' suffering to recognizing shared human experiences and potential for transformation, linking this to a deep empathetic understanding 6 .

6. Information Processing Levels:

   • Tim Scarfe and Yannic Kilcher discuss understanding information processing at different levels: computational theory, representation and algorithm, and hardware implementation. They debate the generality and utility of reinforcement learning in this framework 7 .

These discussions illustrate that understanding is multi-dimensional, covering theoretical, practical, behavioral, and empathetic aspects.