BIOMIMETIC CORTICAL NANOCIRCUITS - Bio-Inspired and Nanoscale Integrated Computing - page 460

Transmitters acting via secondary messengers can have short- or long-term effects
on synaptic junction activation. Altering the conductance of the membrane to ions
based on potential differentials or neurotransmitter concentrations creates im-
portant feedback that is vital to the function of a neuron. Further complicating
transmitter function is the long-term chemical retrograde process that directly or
indirectly modulates transmitter release in the presynaptic terminal, a form of
extremely local feedback.
In Section 17.4, we will describe circuit simulations of carbon nanotube neural
circuits. These circuits display variations in behavior related to neurotransmitter
concentrations and reuptake rates. We demonstrate, via circuit simulations, ion
channel mechanisms. Finally we have produced simple prediction models of
neural size, and mathematical models of neural interconnectivity in order to
predict wireability of the synthetic cortex.
17.3. RELATED WORK
Related work in electronic neural modeling, nanotechnology, and measurement of
brain properties is presented here.
17.3.1. Neural Modeling
Substantial research in related areas of neural modeling using electronics has been
performed, although nanotechnologies are not considered. Schu ¨ ffny et al. provide
a good survey of research up to 1999 [17]. Many researchers focus on specific brain
structures, like the retina, or applications, like image recognition. The goal of a
very small number of projects is construction of an entire artificial brain or cortical
columns consisting of many neurons. Some highly visible research projects [8, 18]
envision synthetic structures built using general purpose processors, specialized
architectures such as cellular automata [19], or asynchronous ARM processors
[18]. Wells [20] proposes a neurocomputer architecture intended to solve the
problems of interconnectivity, variable synaptic weights, and learning. Moravec
[3] has performed optimistic predictions of when inexpensive general purpose
computers will match the human brain in processing power.
The most notable research on electronic neural circuits includes Mead's
artificial retina [21], and continuing work by Boahen (e.g., [1]) and others [22]
on neuromorphic circuits that emulate the behavior of individual neurons. This
significant body of work originated with Mahowald and Mead's pioneering
research [23]. Boahen, who studied with Mead, has also concentrated on visual
processing [7, 24]. Hynna and Boahen report on a circuit that generates a calcium
spike with exact replication of waveforms, and describe incorporation of the
calcium spike circuit in an entire neuron circuit [25]. Some mixed-signal electronic
models close to biological neurons include Liu and Frenzel's spike train neuron
[26], Pan's bipolar neuron [27], Wells' research (e.g., [28]), and Chua's cellular
neural network (CNN) [19, 29]. Chiju et al. extends the CNN work and tests their