There are several studies that estimate the emergent event from the time series of behavior in some organisms. However, they do not focus on the emergent event itself. Our aim is to detect the emergent event from the time series of individual's behavior, focusing on the transition from predictable machinery behavior to purpose-oriented behavior and vice versa. We recorded the behavior of larvae and adults of black larder beetle. To detect the emergent event of the beetle, we defined a forward- and backward-prediction model. In the forward-prediction, the next state in the time series of behavior was interpreted by precedent behavior. In the backward-prediction, the previous state in the time series of behavior was interpreted by subsequent behavior. The time step with conspicuous peak of the co-intensity of errors in the forward- and backward-prediction was regarded as the timing at which the emergent event occurs. At the same time, the time series of states was estimated to determine whether noise was stationary or non-stationary. The attribute of noise was estimated using the Allan variance. The time series of the larvae's velocity of walking showed stationary noise. But in the case of the adults, whole time series contained 1/f noise. And, when time series was divided before and after the detected event, the noise changed from stationary to non-stationary and vice versa. These results suggest that development enables an individual to change the internal mechanism of walk considering the slight change of environment.
ASJC Scopus subject areas