Sampling behavior in a four instrument monitoring task: effects of signal bandwidth and number of events per signal

Abstract

This study investigates the effect of global signal characteristics, such as bandwidth and number of events on human sampling behavior. It also examines whether the influence of these characteristics on sampling strategy is affected by local (actually observed) signal features, such as the degree with which a sampled signal value falls short of an event region and the rate of change. In the present study four independent, numerically displayed signals were used; two different bandwidths and two different event numbers were chosen. To take a sample, subjects had to use a mouse. The mouse key responses were used as an index of sampling. The study demonstrates that both bandwidth and number of events equally affected the distribution of samples over signals. In addition, it shows that global signal characteristics determine sampling behavior less prominently when the attentional demands brought about by the local signal features become of greater importance. This indicates that not predictability as governed by global signal characteristics as such, but rather predictability given certain local signal features is a crucial factor in determining sampling behavior.