EFFECT OF SAMPLING FREQUENCY ON AUTOMATICALLY-GENERATED ACTIVITY AND FREEZING SCORES IN A PAVLOVIAN FEAR-CONDITIONING PREPARATION

Conditioned freezing has long held conceptual importance in behavior analysis, being pivotal in the explanation of anxiety-like behavior. Although initially measured indirectly, through its disruptive effect on operant behavior (conditioned suppression), and later by direct observation, automated techniques of measuring movement have recently become available, which also enable the measurement of conditioned freezing. These video processing techniques allow for the direct and virtually continuous measurement of activity, as compared to the traditional interval sampling approach of direct observation. We examined whether automatically generated freezing and movement scores were equally sensitive to traditional Pavlovian conditioning manipulations, and how this sensitivity was affected by the sampling frequency of the data. Extinction data for 42 mice were collected at a rate of 30 Hz, transformed via re-sampling and analyzed by a generalized linear model to determine the effect size for the presence of the conditioned stimulus for each individual time series under four conditions: high and low resolution raw activity scores and high and low resolution dichotomous freezing scores.The resolution of the data proved to be the most important dimension in estimating local changes in the level of the individual time-series, with activity and freezing scores presenting comparable effect sizes. In contrast with the above, only high-resolution activity measurements proved to be effective in detecting local changes in trends.