The Temporal Patterning of Spontaneous High-Rate Behavior: An Evaluation of Bout Partitioning Methods for Behavior Under Neurotoxicant Challenge

Abstract

Across species, much free-operant behavior exhibits a temporal pattern characterized by bouts of activity and periods of disengagement. Analysis of interevent time distributions has been widely used to characterize bout structure. As an alternative to such methods, we propose a novel bout analysis technique that uses change-point detection to partition a cumulative record into activity epochs directly. The utility of the change-point bout analysis was examined in two studies. The first study directly compared the change-point analysis to the two most widely used bout analysis methods, log-survivor analysis with a biexponential model and log-interval analysis with a Gaussian mixture model. All three methods provided similar descriptions of bout structure for data simulated to follow the standard theoretical model, but the log-interval analysis produced less accurate parameter estimates for several bout structures. When applied to spontaneous wheel running data, the change-point analysis proved as good as or better than the conventional methods. The log-survivor consistently overestimated response rate and the log-interval analysis produced an implausible description of the bout structure. All three methods detected a significant methylmercury-related decrease in within-bout rate, but the change-point analysis also detected a significant reduction in bout length. In the second study, the change-point analysis was used to characterize changes in the bout structure of spontaneous wheel running in BALB/c mice chronically exposed to 0 or 15 ppm methylmercury. Study two also examined the potential role of calcium in methylmercury neurotoxicity using co-treatment with the calcium channel blocker nimodipine. Methylmercury exposure increased mortality rates and substantially decreased total running. The change-point analysis revealed that the decreases in global running were the result of declines in within-bout running rate and the length of response bouts. Nimodipine dose-dependently protected against those effects.