ABSTRACT
Whenever neurobiological data contains two or more streams of events in time (such as action potentials, postsynaptic potentials, discrete stimuli), it is appropriate to seek temporal relationships between the streams. Basically we search for coincidence or near coincidence of events in the several streams; the appropriate mathematical tools are related to correlation. Direct application to the data produces what we will call the “raw” correlation or coincidence structure. However, even completely random and unrelated streams will show some degree of “accidental” coincidence; this level will be modified if two or more streams are influenced by the same modulating
process (stimuli, for example). Thus it is necessary to develop a measure that extracts the excess or deficit coincidence structure, i.e., the deviation from the amount expected by chance and from all known sources of shared influence. Finally, we need a significance measure for this measured deviation.
