Molar vs molecular approaches to reward schedule and serial learning phenomena

In Experiment 1 all rats received identical series of rewarded and nonrewar ded trials in a black runway and in a white runway. A grouping cue, a chang e in runway brightness, was introduced on a rewarded trial that followed ei ther a single nonrewarded trial (Group N1) or four successive nonrewarded t rials (Group N4). Over a series of four nonrewarded trials terminating in r eward, Group N1 ran slower than Group N4 on Trial 2 but faster than Group N 4 on Trial 3. In Experiment 2, slower running occurred when the grouping cu e occurred on Trial 4 of a consistent reward schedule rather than on Trial 4 of a partial reward schedule. These findings were shown to be inconsisten t with three theories that attempt to explain reward schedule data in terms of some overall characteristic of the schedule such as percentage of rewar d (molar theories). The data are consistent with the sequential view that r ecommends decomposing reward schedules into more elementary memory units (a molecular theory). In particular the data demonstrate that the effects of an overall reward schedule on behavior are determined by the more specific reward schedules associated with each memory component of the schedule. Imp ortantly, the findings suggest that reward schedule investigations and seri al learning investigations are, theoretically speaking, identical. Accordin gly, the findings strongly discourage the common practice of reward schedul e theories ignoring serial learning data and of serial learning theories ig noring reward schedule data. (C) 2001 Academic Press.