Detecting synfire chains in parallel spike data

► A Synfire Chain is a theoretical model for portions of the nervous system. ► Synfire Chains have many computational properties appropriate for the real nervous system. ► Experimental verification of the model requires simultaneous recording of at least 100 neurons. ► We describe three data analysis tools that produce characteristic signatures if a synfire chain is present. The synfire chain model of brain organization has received much theoretical attention since its introduction (Abeles, 1982, 1991). However there has been no convincing experimental demonstration of synfire chains due partly to limitations of recording technology but also due to lack of appropriate analytic methods for large scale recordings of parallel spike trains. We have previously published one such method based on intersection of the neural populations active at two different times (Schrader et al., 2008). In the present paper we extend this analysis to deal with higher firing rates and noise levels, and develop two additional tools based on properties of repeating firing patterns. All three measures show characteristic signatures if synfire chains underlie the recorded data. However we demonstrate that the detection of repeating firing patterns alone (as used in several papers) is not enough to infer the presence of synfire chains. Positive results from all three measures are needed.