Bidirectional interference in an A-B, C-B paradigm

Abstract

In Exp. I, two transfer paradigms (A-B, C-B and A-B, C-D) were combined factorially with two levels of relearning direction (A-B and B-A), and an S-R matching task was used during relearning. Relearning performance was affected only by transfer paradigm ( p < .05); the C-B paradigm was conducive to RI, the magnitude of which was independent of relearning direction. C-B interpolated learning tended to be enhanced on initial trials owing to positive transfer of response learning, but retarded on later trials owing to bidirectional negative transfer of associative learning. Experiment II eliminated the positive transfer effect by employing an S-R matching task during transfer; as a result, the negative transfer effect was more pronounced and was evident on initial trials. The data support the concept of associative symmetry and Martin's (1965) multistage analysis of transfer.