Interference in Memory: A Matter of Encoding, Not Retrieval

The study of forgetting has attracted increasing interest in recent years, with the advancement of neurobiologically-inspired theories. According to our Representation Theory of Forgetting, forgetting may occur due to either interference or delay, depending on the underlying memory representation and the brain structures supporting it. Memories dependent on the hippocampus, whose retrieval relies on recollection, are relatively protected from interference, but prone to delay-dependent forgetting.  Conversely, memories dependent on the extra-hippocampal regions, whose retrieval is driven primarily by familiarity, are forgotten mostly due to interference and are less susceptible to delay-dependent forgetting. Here we focus on forgetting due to interference. The major goal of the current study was to examine whether interference exerts its effects primarily at encoding, retrieval or both phases. Interference was defined as an overlap between representations, and thus manipulated as the number of items similar to the targets. As hypothesized, interference predominantly impaired memory for familiarity-based representations. Most importantly, the degree of interference was found to exert an effect at encoding. In contrast, increasing the number of similar lures at retrieval had no detrimental effect on memory accuracy. In a follow-up experiment which focused on interference at encoding, we found that interference caused by exposure to multiple items similar to the target hinders within-category discrimination, while improving across-category discrimination.  Taken together, our results indicate that interference caused by the presentation of similar information at encoding results in overlapping representations of the similar information. This, in turn, hampers memory specifically of within-category discriminations. Our results may have far-reaching implications for research and practice of educational assessment by showing that while memory accuracy is greatly affected by the number of similar items at study, it is not affected by the number of distractors presented at test.