Regression analysis of non-stationary discharges in neurons; Adaptation in the electrosensory afferent of dogfish

Abstract

Interspike interval histograms, as usually regarded for the estimation of statistical variabilities in neuronal spike trains, were applied to non-stationary dynamic responses of a PD receptor. Sliding mean values were introduced describing the average receptor response on defined, recurrent stimuli; mean spike frequencies and interspike intervals were computed a) for fixed sequential analysis periods (of e.g. 500 ms), b) for analysis periods shifted by every consecutive interspike interval (thus the number of spikes being constant), and c) by fitting the dynamic responses for suitable analytic functions (e.g. exponential functions). With these methods variabilities in the non-stationary neuronal impulse patterns were investigated for electrosensory PD afferents in Lorenzinian ampulla of dogfish (Scyliorhinus canicula) with electric stimuli up to 50 nA and defined temperatures between 7° C and 25° C. In this temperature range all investigated ampullae were spontaneously active, the irregularities in neuronal discharges and averaged spike frequencies depended strongly on temperature, the latter showing maxima between 13° C and 19° C. In preparations with small disturbances we generally found static interspike interval histograms following approximatively a Gaussian distribution. The same was true for the momentary spike frequency and its deviation during the dynamic response to given electrical stimuli. A suprathreshold rectangular current (e.g.-0.5 nA) led to a marked but transient synchronisation in spike generation; the higher the stimulus strength, the smaller the standard deviation (s.d.) from mean spike frequency in the beginning of the dynamic response; during adaptation the s.d. increased up to that of the static response frequency. Relating, however, s.d. for different currents, times, and temperatures to the corresponding mean spike frequency led to fairly constant coefficients of variation; s.d. was approximatively a linear function of the sliding mean value even in the dynamic response of the electroreceptor (scaling).