Max Planck Institute for Dynamics and Self-Organization -- Department for Nonlinear Dynamics and Network Dynamics Group
Personal tools
Log in

BCCN/BFNT AG-Seminar

Tuesday, 06.07.2010 17 c.t.

What makes brains efficient?

by Prof. Dr. Simon Laughlin
from Department of Zoology, University of Cambridge, UK

Contact person: Fred Wolf

Location

Seminarraum Haus 2, 4. Stock (Bunsenstr.)

Abstract

Brains appear to be remarkably efficient. Our 1.5 kg brain consumes 20 W of power, performs as many operations per second as a 2.2 MW supercomputer, and is both autonomous and intelligent. The tiny fruit fly brain supports “fly by wire” in complicated environments, pattern recognition and learning and, by weighing up context and life history, enables the fly to gamble expertly in the casino of life. How can brains do much with little? To determine the efficiency of a device we must know its purpose, its mechanisms and its cost. To see if it is designed to be efficient (either by an engineer or by Natural Selection), we must understand how physical constraints on mechanism set limits and determine trade-offs between cost and capability. To meet these requirements our studies of efficiency have focused on discrete signals (e.g. action potentials), the generic properties of neural codes, and information transmission and processing in well-defined retinal circuits where, because function is known, performance can be measured and related to costs. Using experiment and theory we find that efficiency is achieved by trading the costs of materials and energy consumption against speed and accuracy so as reduce investment and effort to the minimum required for adequate function. This suggests that slow neurons and circuits with poor representational capacity make brains efficient.

back to overview