|
Abstract:
|
Detecting dependencies in the spatio-temporal pattern of neuronal spiking activity is challenging, in part because many different types of dependencies are confounded in the data. Classical permutation tests can often be used to distinguish time-locked from non-time-locked dependencies. These methods, however, do not easily disambiguate interactions on multiple time scales and do not extend to situations with single or non-repeatable samples. We have been developing techniques, called jitter methods, designed to investigate the time scale of arbitrary spatio-temporal dependencies in neuronal spiking activity. Jitter methods are analogous to classical permutation tests except that the permutations are over spike times, instead of trial labels. Jitter methods perform well in a variety of settings where dependencies on multiple time scales are confounded.
|
