Monday, March 6th 12:30-2pm — ACE teleseminar room
Lecture Series Organised by Simon Penny
This talk co-sponsored by
UCI department of Anatomy and Neurobiology
How does one go about building artificial devices that are capable of growing new structures that create novel functions in an open-ended manner? How does one organize a system that can continuously construct itself? How do we construct autonomous systems with intentions of their own, i.e. whose embedded purposes steer their behavior? How do we construct neural networks that can create new internal signals for themselves that can be imbued with meaning?
We will discuss various primitive artificial mixed analog-digital systems (von Neumann’s kinematic self-reproducing automaton, Ashby’s homeostat, Pask’s electrochemical assemblage), their underlying functional principles, and their similarities to biological systems. We will also consider the brain as a self-organizing mixed analog-digital cybernetic system in which feedback loops link internal and external events to create anticipations of consequences (meanings) and to steer behavior in accordance with internal goals (pragmatics). Finally we will outline an alternative vision of neural networks in which the basic signals are temporal patterns of spikes. Such temporal coding makes possible signal multiplexing, broadcast and selective reception, statistical nonlocal storage of temporal memory traces, and creation of new signal types, topologically more like radio or internet than telegraph network or phone switchboard.