Stream 2c: Evolutionary Learning

There is not enough information in the human genome to hard-code the representations used by the brain that support flexible cognition.  Instead, such representations and resulting cognitive competencies emerge from interactions with the environment, given an appropriate cognitive architecture that supports the learning process.  Take for example, the “simple cells” in area V1 of the primary visual cortex that respond to visual edges of a particular orientation - rather than being hard-coded into the system, it is believed that they reflect statistical regularities in the real world that can be learned through experience.

To effectively learn the knowledge that is required to function in the real-world requires both the appropriate learning algorithms, as well as the appropriate architecture and local connectivity patterns of neurons.  Hand-generating such structures may be as difficult as hand-coding the knowledge to be learned itself.  In biological systems, such structures are specified in the genome, being generated through evolution.  Likewise, simulated evolution may be useful in generating appropriate architectures that support effective learning in artificially intelligent systems.  The final research stream of this project will aim to investigate this issue.

Home - Stream 1 - Stream 2a - Stream 2b - Stream 2c