Hierarchical Concurrency and the Brain

Anthony S. Maida

Center for Advanced Computer Studies

Institute of Cognitive Science

University of Louisiana at Lafayette

 

Abstract

The brain is a massively parallel computer implemented on special purpose biological hardware. The computing element in the brain is a neuron. The neocortex (most recently evolved part of the brain) has 28 billion neurons, in comparison to the 42 million transistors in the early Pentium IV processor. There are perhaps twice as many neurons in the cerebellum. Some neurons in the cerebellum receive inputs from as many as 200,000 neurons. In the neocortex, fan in is approximately equal to fan out with a branching factor of about 10,000.

The above describes the low-level parallelism in the brain. The neocortex is also exhibits hierarchical parallelism. Neurons form minicolumns. Minicolumns form macrocolumns. Macrocolumns form cortical sheets. Cortical sheets interact to perform computations that we call thinking.

This presentation describes this concurrency and hierarchical organization with an eye toward developing large scale brain simulations on the new LITE facility.