Hierarchical organization in serial pattern learning

Theorists have long hypothesized how organisms represent the organized sequences of stimuli that they encounter. Such models often posit that the items that are encountered are not coded individually, but instead that the organism constructs an abstract representation of the sequence based upon their experience. Later, this representation can be drawn upon to reconstruct the entire sequence. However, the nature of that representation continues to be debated. Through two experiments with human undergraduates in a pattern production paradigm, we consider whether one popular model, the hierarchical model (Restle, 1972 and Restle and Brown, 1970), accurately describes such learning. The hierarchical model predicts that humans should form an abstract representation of the sequence with which they are presented that is the simplest possible in order to reduce memory load. It posits a nested organization of relationships in which the highest-order rules relate the largest number of pattern elements and sets of elements while the lower-order rules are nested within this higher-order structure; the lowest-order rules relate individual pattern elements. However, our results indicate that participants did not abstract the simplest representation of the sequence available, contradicting the prediction of the hierarchical model. This suggests that the hierarchical model does not fully account for the learning of patterned sequences in humans.