Variable Training for a Novel Instance in a Dynamic Environment

Abstract

The study used an “open” or dynamic environmental task to determine whether the acquisition of novel responses outside the range of the subjects' past experience would be facilitated by variability of training. This was found to be the case for a coincident-timing task. Subjects with high stimulus variability during training showed improved transfer to a novel task. A generalization gradient was obtained; the further the novel conditions were from the training conditions, the poorer was the transfer. However, generalization decrement was attenuated with variable training.