Colloquium

The Scalable Architecture of the Brain and an Underlying Universal Scaling Law

Charles F. Stevens  
The Salk Institute
Wednesday, March 21, 2007
Higgins 310, 4 pm

All vertebrate brains conform to one basic design – with elaborations – and this design has the property that the brain’s computing power can be increased by simply making the brain, or parts of it, larger. I am interested in the design principles that endow vertebrate neural circuits with this scalable architecture.

Information is spread over regions of the brain by axon arbors, and this information is sampled by dendritic arbors. These arbors are statistically self-similar and follow a universal scaling law that relates arbor length to the size of the region over which information is distributed or sampled (the arbor’s ‘territory’). The exponent in this scaling law is the result of the principle that potential for making a particular neural circuit is independent of scale.