Comparative analysis of measures of association between spike trains

This paper discusses the problem of detecting functional connections btween pairs of neurones. Two specific measures of association are considered in addition to the commonly used peristimulus time histogram (PSTH) approach. The chosen methods involve use of CUSUM and coherence based measures and both are applied to the same data sets involving pairs of spontaneously discharging fusimotor neurones. In the cases examined the coherence measure was found to be the most sensitive measure of association. It also has the advantage of the other methods in providing an absolute measure of the degree of linear dependednce on a scale of 0 to 1