Friday, May 14, 2010

Contingency and learning.

Contingencies, the mixture of predictability and unpredictability in the occurrence of events, has an important significance in nurturing our brains. If everything follows some already known rules, there is nothing to learn any more. If, on the other hand, the events occur in a random manner, as in the case of thrown dices, there isn't anything to learn either, except for the realization that the statistics of dice indeed exhibits randomness. Any "learning" beyond that would be due to a gambler's fallacy.

Thus, it is always the case that a mixture of predicable and unpredictable elements provides an opportunity for learning. It is not that the unpredictable elements monotonously decreases as the learning progresses. It is rather that learning new regularities leads to a structuring of the world, in which newly unpredictable aspects of events emerge in our cognition. Thus, the brain is playing an incessant game of pursuit and chase, in which the unpredictable is never diminished to nothingness.


Yuzu said...

I try to jump into the ocean of uncertainties,I think. But it is quite hard to swim. And I am so slow.
I almost cry everyday. But I have read your journal ,books and words.

Jordan said...

Prof Mogi,

This sounds somewhat like what statistical modeling would require.
For example linear regression would do its best to fist a line (a pattern) while using least squares estimation for instance in order to reduce the random error as much as possible.