Friday, May 19, 2006

Your Brain is not a Computer – Part 3

In the game of chess, you don’t have to worry about your castle suddenly assassinating your own bishop, or your queen having an affair with one of the opponent’s knights, or your pawns revolting and refusing to move forward. In chess, you don’t learn to improvise, you just learn patterns. To master chess is no more extraordinary than a kid who masters all the levels of a video game in a week or a person that memorizes all the religious scriptures of a holy book. Chess is for robots. Improvisation is for humans.

Deep Blue is held up as a pinnacle of brain modeling. But what do you think would happen, if Deep Blue were engaged in a chess match and, several turns in, you told Deep Blue, “The opponent’s king has been removed from the board because he died of a syphilitic infection and the queen has now seized the throne. The queen can still move in any direction but is limited to five squares. The bishop still moves diagonally but has been granted the movement of a knight as an option. Finish the game.”

How well could Deep Blue improvise on this information? How long – how many games - would it take for Deep Blue to rebuild a competitive matrix? I don’t know the answer. I am only suggesting that, with a human, there is a slight adjustment and not a major reprogramming.
MRI image with regions of interest (ROIs) from BrainMiner

Part 1
Part 2