Neither Rescorla and Wagner nor SOCR Predict the Double Blocking Effect

Abstract

In a traditional blocking experiment, A+ trials precede AX+ trials, and responding to X at Test is weak relative to controls. Using the allergist task, the present study investigated the effect of an additional blocking cue presented in compound (i.e., ABX+, where B is also a previously trained excitor). The study found evidence of a double blocking effect, a proposed associative learning effect in which blocking with two blocking cues results in increased behavioral control by the target cue relative to blocking with a single blocking cue. In Phase 1 participants received A+/B+ training, in Phase 2 participants received ABX+ trials. Relative to controls, responding to X was greater in the double blocking group than in the traditional single blocking group. The Sometimes-Competing Retrieval (SOCR) model ostensibly should be able to explain double blocking through second-order comparator processes; however, a simulation analysis determined that SOCR was unable to satisfactorily explain the effect or fit the present data. Moreover, Rescorla-Wagner did no better than SOCR in fitting double blocking. The double blocking effect thus presents a challenge yet to be surmounted by a model.