Data Skeptic Memory in Chess
Feb 12, 2024
The podcast discusses perception and memory in chess, exploring how chess players recall positions on the board. It covers topics such as cheating with chess engines, chunking strategies, and studying chess cognition through eye tracking and MRI. The podcast also delves into algorithms in checkers and chess, the capabilities of AI in games, and differences between Deep Blue and AlphaGo. It concludes with a discussion on the challenges of playing Kriegspiel, a version of chess with hidden pieces.
AI Snips
Chapters
Transcript
Episode notes
Becky's Animal Cognition Training
- Becky Hansis-O'Neil studies bumblebees and tarantulas to understand learning and cognition.
- She trains tarantulas to get in buckets for heart rate measurement, showing animal memory in practice.
Grandmaster Recall Depends on Patterns
- Grandmasters recall meaningful chess positions far better than beginners.
- When positions are randomized, all players recall similarly, showing chunking depends on pattern recognition.
Chunking Simplifies Chess Memory
- Chunking condenses chessboard memory into meaningful groups rather than individual pieces.
- Grandmasters recognize complex structures like openings as single memory chunks.