Analog Computation to Model Responses Based on Linear Integration, Modifiable Synapses, and Nonlinear Trigger Zones

Abstract

Up to this time my approach had been to identify an experimental domain of linear dynamics by means of paired-shock stimulation in order to show that proportionality and additivity held for the superimposed responses in certain ranges of the stimulus pulse interval and intensity (Biedenbach and Freeman 1965). Using measurements of cortical impulse responses by curve fitting, I constructed a series of linear dynamic models and evaluated the coefficients, extending their ranges by varying the coefficients in piecewise linear approximations. In the 1960s the use of analog hardware to model nonlinear systems had become widespread, so it seemed worthwhile to construct a network model, using operational amplifiers to simulate the linear integration performed by the dendrites of cortical populations, and diodes to simulate the static nonlinearity at the trigger zones (Figure 7 in Chapter 7), treating pulse densities as continuous variables in time and amplitude. The circuit connections were the specified by the KII model. My aim was to simulate the changes in the waveforms of evoked potentials and the root loci with increasing stimulus intensity in Mode 1, and then to simulate the patterns of change and root loci in Mode 2 (Chapter 6).