Abstarct learning

Human learning is not just a pattern recognition filter, but involves forming and fine-tuning abstract models of aspects of the world. We have the ability to discover general rules behind specific examples, and to apply these rules to new examples.

  • We can add two large numbers, while crows can do that up to a sum of about 10.
  • Most children discover at age 5-6 that each natural number n has a successor n+1 and infer that the sequance of natural numbers is infinite
  • George Cantor proved that infinities come in infinitely many sizes.

    • George Cantor

    Among primates humans are the only ones who can detect patterns in arrangements of abstract symbols efficiently. It takes 4 year old human child on average 5 trials to learn how to reverse a sequence of 4 symbols: CDBA -> ABDC. For a chimpanzee it takes tens of thousands of trials.


    Jerry Fodor postulated that human thinking consists of a language of thought, a system of mental representations that have a combinatorial syntax and semantics (1975).

    PreviousNext