Retuning conceptual-lexical access: does interference promote more robust learning?

Abstract

Generative activity can interfere with later retrieval of related concepts and words. This semantic interference has been ascribed to long-term adaptive learning mechanisms that promote access to selected representations and hinder access to coactivated competitors. Some accounts in the memory literature posit that the punishment doled out to coactivated competitors may actually serve as a catalyst for more robust relearning in comparison to never-punished controls. The present work investigates whether this claim also applies to lexical access by assessing how adaptive learning processes unfold during generation of taxonomically related words in each phase of a modified retrieval practice and relearning design. In each phase, there was cumulative semantic interference across category member ordinal position, demonstrating that interference accrues with each related retrieval. The final assessment phase showed persistent semantic interference in the absence of relearning, but naming was equivalent for activated and control conditions that were subjected to relearning. These results provide a detailed window into the temporal dynamics of learning and relearning in conceptual-lexical access but suggest that enhanced relearning chiefly applies to novel learning episodes and not well learned conceptual-lexical links.