A large body of literature agrees that accessing a target memory appears to trigger a difference-of-Gaussian memory activation pulse under which the target representation is activated and categorically flanking items are suppressed and forgotten. The nature of the underlying forgetting mechanism is far from settled, with support for several theories of forgetting. Here we argue the debate is partly fueled by different forgetting mechanisms underlying the forgetting of different memoranda. We capitalized on the unique aspect of the recognition-induced forgetting paradigm to test forgetting of both pictures and words in identical recognition-practice and restudy tasks. We found that memory for pictures and words followed different patterns of forgetting. Specifically, forgetting was retrieval specific for words, in that forgetting occurred only when words were recognized, and not when words were merely restudied. However, forgetting was not retrieval specific for pictures, in that forgetting occurred both when pictures were recognized as well as restudied. Further, patterns of forgetting operated along different category-level groupings for pictures and words. Words grouped along the superordinate level were susceptible to forgetting but pictures were not. The strength of this design is the ability to directly compare forgetting for different memoranda, establishing that patterns of forgetting are modality specific. These findings demonstrate that the mechanisms underlying forgetting may differ as a function of the particular memoranda, emphasizing the need for examining forgetting in long-term memory across modalities.

Recognition-induced forgetting is a within-category forgetting effect that results from accessing memory representations. Advantages of this paradigm include the possibility of testing the memory of young children using visual objects before they can read, the testing of multiple types of stimuli, and use with animal models. Yet it is unknown whether just episodic memory tasks (Have you seen this before?) or also semantic memory tasks (Is this bigger than a loaf of bread?) will lead to this forgetting effect. This distinction will be critical in establishing a model of recognition-induced forgetting. Here, we implemented a design in which both these tasks were used in the same experiment to determine which was leading to recognition-induced forgetting. We found that episodic memory tasks, but not semantic memory tasks, created within-category forgetting. These results show that the difference-of-Gaussian forgetting function of recognition-induced forgetting is triggered by episodic memory tasks and is not driven by the same underlying memory signal as semantic memory.

Previous research from our lab has shown that recognizing an object stored in visual long-term memory leads to the forgetting of related objects. Here we ask whether context, an integral aspect to modern models of memory, plays a role in induced forgetting. We manipulated the activated context at test, both externally (e.g., changes in testing room) and internally (e.g., 1 hr and 24 hr later). We found that only interfering with the ability to internally reinstate context after 24 hr eliminated induced forgetting. Thus, we demonstrate that mental context reinstatement plays a role in induced forgetting and specify that models of memory should incorporate internal context reinstatement as an underlying factor of forgetting. We also propose a process model of induced forgetting, discuss limitations of laboratory-based memory tasks, and offer a new term, induced suppression, to collectively describe this robust phenomenon. Public Significance StatementContext plays a substantial role in many models of recognition memory. In 1 such model, forgetting is entirely due to context. The present article sought to test the hypothesis that changes in contextual retrieval underlie forgetting in laboratory-induced forgetting of pictures of real-world objects. We manipulated both spatial and temporal context and found that forgetting survived changes in space, but not changes in time. Induced forgetting was brought back online when the practice phase, which induces forgetting, was also delayed 1 hr and 24 hr. The elimination of induced forgetting after 24 hr appears to be accounted for by models of forgetting that incorporate a role of context, and difficult to account for otherwise. We conclude by presenting a process model of recognition-induced forgetting and suggest that induced forgetting is better characterized as induced suppression.