TY - JOUR
T1 - A parametric examination of VLSI-based neuronal models of cyclic and reciprocal inhibition
AU - Wolpert, Seth
N1 - Funding Information:
Manuscript received November 30, 1994, revised June 6, 1996. Developmental work on the IC-based Neuromime and neural oscillator networks was supported by NSF Grants MIP-9210945 and EID-920039 Design and testing of the VLSI Neuromme and oscillatoi networks was supported by the NSF under Grants MIP-9210945, CDA-9320440, and EID- 920039. The author is with the School of Science, Engineering, and Technology, Pennsylvania State University, 777 W. Hamsburg Pike, Middletown, PA 170574898 USA (e-mal: sxw33@psuvm.psu.edu). Publisher Item Identifier S 0018-9294(96)07844-5.
PY - 1996/12
Y1 - 1996/12
N2 - As a prelude to a silicon implementation of a live locomotory network, the phenomena of tonic excitation in single cells, reciprocal inhibition in pairs of cells, and recurrent cyclic inhibition in rings of five cells were recreated and subjected to parametric tests of oscillatory range and stability. Various networks were constructed from comprehensive very large scale integration (VLSI)-based artificial neurons and their parametric stability with respect to cellular threshold, refraction, and synaptic weight observed. Circuit tests demonstrated that all three oscillator topologies operated over a broad range of cellular and network frequencies. It was also noted that cells of reciprocal oscillators must possess some measure of short-term synaptic plasticity while those in the cyclically inhibited networks did not. This suggests that the two oscillator types utilize different temporal mechanisms. In parametric tests, tonic cells were sensitive to threshold, refraction, and synaptic weight. Cyclic networks were-found to be sensitive to cell threshold, yet less so to refraction. Conversely, the reciprocal circuits were found to be sensitive to refraction, yet less so to cell threshold. This complementary relationship suggests a stability advantage for biological oscillatory networks that incorporate both types.
AB - As a prelude to a silicon implementation of a live locomotory network, the phenomena of tonic excitation in single cells, reciprocal inhibition in pairs of cells, and recurrent cyclic inhibition in rings of five cells were recreated and subjected to parametric tests of oscillatory range and stability. Various networks were constructed from comprehensive very large scale integration (VLSI)-based artificial neurons and their parametric stability with respect to cellular threshold, refraction, and synaptic weight observed. Circuit tests demonstrated that all three oscillator topologies operated over a broad range of cellular and network frequencies. It was also noted that cells of reciprocal oscillators must possess some measure of short-term synaptic plasticity while those in the cyclically inhibited networks did not. This suggests that the two oscillator types utilize different temporal mechanisms. In parametric tests, tonic cells were sensitive to threshold, refraction, and synaptic weight. Cyclic networks were-found to be sensitive to cell threshold, yet less so to refraction. Conversely, the reciprocal circuits were found to be sensitive to refraction, yet less so to cell threshold. This complementary relationship suggests a stability advantage for biological oscillatory networks that incorporate both types.
UR - http://www.scopus.com/inward/record.url?scp=0030560946&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=0030560946&partnerID=8YFLogxK
U2 - 10.1109/10.544340
DO - 10.1109/10.544340
M3 - Article
C2 - 9214835
AN - SCOPUS:0030560946
VL - 43
SP - 1164
EP - 1175
JO - IRE transactions on medical electronics
JF - IRE transactions on medical electronics
SN - 0018-9294
IS - 12
ER -