Failure of large doses of ethanol to eliminate Markovian serial dependencies in neuronal spike train intervals.

The effect of drug-induced state changes on serial ordering of interaction potential intervals was evaluated by determining if the intervals were generated by a Markov process, and if they were, determining the Markov level (the degree of serial ordering) of the process before and after the injection of an intoxicating dose of ethanol. When categorizing intervals as short, medium, or long, a Markov process of level 4 (i.e., the process has a "memory" for 4 preceding adjacent intervals) was noted for most neurons; this level of serial ordering remained constant after ethanol injections. Many of the specific clusters that made the major contributions to the statistical significance before ethanol did not contribute after ethanol and a new set of clusters made the major contribution. This suggests that neurons may process intervals as "clusters" and the "cluster-length" may be a basic function of the neuron and not dependent on the general state of the organism. This methodology allows us to determine the "cluster-length" of the processor and detect and isolate the specific "clusters" that may be important state-sensitive "information carriers" (signal) in the nervous system.