DISTRIBUTION-FREE GRAPHICAL AND STATISTICAL-ANALYSIS OF SERIAL DEPENDENCE IN NEURONAL SPIKE TRAINS

Two-dimensional 'joint' interval distributions of sequential interspik e intervals (ISIs) are a commonly used tool in neuronal spike train an alysis. We present and evaluate here a modification of the joint inter val plot using ranked ISIs, This modification provides clearer graphic al evaluation of serial dependence in ISI sequences, a distribution-fr ee basis for isolating changes in serial dependence across experimenta l treatments from changes in ISI distributions, and a basis for unambi guous statistical tests of serial dependence and stationarity, To vali date this method and illustrate the advantages of its use we have appl ied it to both single-neuron spike trains recorded from cultured mamma lian spinal cord neurons and artificial spike trains generated by stoc hastic models with defined burst envelopes and serial dependencies.