The heterogeneous sampling of behavioral states by freely moving animals hinders our ability to relate neuronal firing rates to behavioral variables by introducing dependencies between them. We specifically consider the animal's location and orientation, although our analyses may generalize to other behavioral variables, such as speed of movement. A maximum-likelihood approach is presented for producing estimates of the separate histograms relating firing rate to multiple independent causes. Examples show that the method can be used to avoid the artifactual behavioral correlates of place and head direction-cell firing produced by standard analyses; to characterize the independent influences of both location and orientation in a third cell type (Cacucci et al., 2004); and to demonstrate the location-independence of the directional firing of head-direction cells.
(c) 2004 Wiley-Liss, Inc.