A dorsal spinal neural network in cat. III. Dynamic nonlinear analysis of responses to random stimulation of single type 1 cutaneous input fibers.

The input/output characteristics of a subset of dorsal horn neurons in laminae 3 and 4 [( L3,4:SA1,X], see INTRODUCTION; output cells) of cat have previously been examined in the resting unperturbed condition using single or paired input pulses introduced once every three seconds on single slowly adapting type 1 (SA1) cutaneous mechanoreceptor afferent fibers (1, 27, 28). The present study extends this description to the dynamic condition by use of a random-stimulation method developed for the characterization of multiport pulse-input/pulse-output nonlinear systems. A total of 58 SA1 receptor input channels to 29 [L3,4:SA1,X] network output cells were tested individually in 15 spinal cats with several random train stimuli of differing mean input rates [5, 10, 20, 30, 50 pulses per second (pps)]. Simultaneous stimulation of two input channels with independent random trains was performed in 16 units. In each case, zero-, first-, and second-order descriptions of network behavior were obtained; the second-order characteristics of interest were expressed in the form of excitability functions, which are directly comparable with those obtained from condition-test results. Preliminary testing with multiple input pulses suggested that, in addition to the strong second-order effects previously identified, third- and higher-order nonlinearities and effects with long time constants could generate significant rate effects. Nonetheless, first-order response characteristics obtained in the dynamic condition at the lowest mean input rate used (5 pps) were in each case qualitatively identical, though slightly smaller in magnitude, to the poststimulus time histograms (PST) obtained in the unperturbed condition. Second-order excitability functions were generally, but not always, similar to condition-test results in the eight cases in which comparisons were made. Furthermore, use of a complete second-order characterization to predict the output response to a different random input in five cases resulted in an average correlation with the observed output that was a 50% improvement over the linear model predictions. These results indicate strong second-order and weaker higher-order nonlinearities in the [L3,4:SA1,X] network. Three classes of channel-specific second-order excitability characteristics were identified into which the previous descriptions (28) can be incorporated. The general pattern was initial facilitation followed by inhibition. This was observed for both the early and late response components in about half the channels (class I).(ABSTRACT TRUNCATED AT 400 WORDS)