Retrieval-induced forgetting (RIF), demonstrated by Michael Anderson, Robert Bjork, and Elizabeth Bjork (1994), is the finding that practicing retrieval of some items from a studied category impairs later recall of related but unpracticed items from the same category. If you study fruit-orange, fruit-banana, fruit-apple, fruit-grape, and then practice retrieving orange and banana, your later recall of apple and grape is worse than if you had not practiced any fruits at all. Retrieval helps the practiced items but hurts the unpracticed competitors.
The Retrieval Practice Paradigm
The standard paradigm has three phases. In the study phase, participants learn category-exemplar pairs (e.g., fruit-orange, fruit-banana, animal-horse, animal-dog). In the retrieval practice phase, half the items from half the categories are practiced (fruit-or____). In the final test, all items are tested. The key finding is that unpracticed items from practiced categories (Rp- items: fruit-apple, fruit-grape) are recalled at a lower rate than items from entirely unpracticed categories (Nrp items: animal-horse, animal-dog).
Nrp items (baseline control): ~50% recall
Rp- items (related unpracticed): ~35% recall
RIF = Nrp − Rp- (typically 10-20 percentage points)
Inhibitory Account
Anderson proposed that RIF results from an inhibitory mechanism: when retrieving orange in response to "fruit-or____", competing items (apple, grape) are activated and must be suppressed to successfully select the target. This suppression reduces the accessibility of the competing items, producing forgetting. Key evidence for inhibition (as opposed to mere competition or blocking) includes the finding that RIF is cue-independent — it occurs even when competitors are tested with novel cues that bypass the original category cue.
RIF may serve an adaptive function by keeping the memory system efficient. If retrieving a memory strengthened not only the target but also all related memories, the system would become increasingly cluttered with competing associations, making future retrieval more difficult. By inhibiting competitors during retrieval, the system ensures that the most frequently needed memories become progressively more accessible while less-needed competitors fade. This "sharpening" of memory through selective retrieval is an active, ongoing process.
Boundary Conditions
RIF is modulated by several factors. It is greater when competitors are strong (well-learned) — consistent with the idea that strong competitors require more inhibition. It is reduced or eliminated when integration strategies are used (encoding items as parts of an integrated scene). It does not occur for all types of memory tests — some studies find RIF only for recall, not recognition. And emotional memories may be resistant to RIF, potentially explaining the persistence of traumatic memories.
Practical Implications
RIF has implications for eyewitness memory (repeated questioning about some details may impair memory for other details), education (testing some material may cause forgetting of related unstudied material), and clinical applications (retrieval-based approaches to weakening maladaptive memories). Understanding when retrieval helps and when it hurts is essential for designing effective learning and assessment strategies.