BIOMIMETIC CORTICAL NANOCIRCUITS - Bio-Inspired and Nanoscale Integrated Computing - page 457

have short- or long-term effects on synaptic junction activation. The activation
probability of a given synaptic junction is up- or down-regulated by the amount
and timing of presynaptic and postsynaptic activity. Neurotransmitters must be
present in sufficient amounts to develop post-synaptic potentials (PSPs), and the
concentration of transmitters released can affect both the height and duration of
the PSP [9], phenomena that we model.
17.2. CHAPTER OVERVIEW
Our long-term objective is the development of the technology and methodology to
implement a synthetic cortex. While there are hundreds of brain structures that
have somewhat different anatomical characteristics [9], this research focuses on
cortical structures because the density and complexity of their interconnections
poses argueably the greatest neural modeling challenge. The cortex and cortical
neurons exhibit many morphological variations that we will investigate. These
neural circuits should accurately model inter- and intra-cellular mechanisms that
neuroscientists believe are important to neural processing (e.g., the effects of
neurotransmitter concentrations on cellular communication).
Future systems using these circuits could be used to construct large-scale
cortical structures. Because this goal is in the distant future, while working toward
the goal we find it necessary to answer to one simple question: When is technology
going to progress sufficiently to be able to construct a synthetic cortex of
reasonable size and cost that exhibits almost real time behavior? We believe
that such a complex nanotechnological synthetic cortex will eventually be
possible. This chapter describes research results in two areas:
the design and simulation of biomimetic neural nanoelectronic circuits
using future technologies, and
the prediction of the possibilities for three-dimensional interconnectivity of
CMOS neural circuit dies using flip-chip technology.
The emphasis in this chapter is on the first objective: circuits modeling cortical
synaptic structures, designed and simulated using carbon nanotube transistor
SPICE models provided by Wong [10] based on carbon nanotubes fabricated by
Zhou [11, 12]. These cortical structures are being designed by us using custom
biomimetic circuits, sometimes called neuromorphic circuits [1], based on carbon
nanotube transistors. Carbon nanotubes are widely studied nanotechnology
materials that have the potential to support three-dimensional circuit structures.
The custom neural circuits that we present here use advanced nanotechnologies,
are programmable at a very high level (virtually software free), integrate memory
and processing, and execute in parallel in a natural manner. To date, we have
designed and simulated a nanoelectronic neural synapse using carbon nanotube
models, a revolutionary change to the underlying technology.