Retroactive inhibition as a function of the locus of categorized response sets.

The effects which list-level taxonomic attributes should have upon retroactive inhibition (RI) producing mechanisms in both response-set interference theory (Postman & Stark, 1969) and Petrich's (1975) retrieval model are discussed. A test of the theories was performed by manipulating the presence (1) or absence (0) of categorized response sets in both original learning (OL) and interpolated learning IIL), and comparing their predictions concerning the relative degree of RI expected in the four conditions (0-0, 0-1, 1-0, 1-1). In line with Petrich's predictions, OL categorization significantly reduced RI, IL categorization significantly increased RI, and 0-1 produced significantly greater RI than 1-0. Moreover, the obtained ranking of the means for the experimental groups in degree of RI (0-1>0-0>1-0 ≅ 1-1) was in close agreement with the ranking predicted by Petrich's (1975) model. The results provide support for Petrich's (1975) retrieval model and pose explanatory problems for response-set interference theory (Postman & Stark, 1969).