Contextual gating of memory retrieval.

In two experiments, 3-month-old infants learned to move a crib mobile (the cue) in the presence of a distinctive crib bumper (the context) by operant kicking. In Experiment 1A, infants were trained for 2 days and tested either 1, 3, or 5 days later with one of four same/different cue/context combinations. After all delays, infants tested with the original cue and context exhibited excellent retention, and those tested with a different cue and context exhibited none. Changing the context but not the cue disrupted retention after 3 and 5 days but not after 1 day; in contrast, changing the cue but not the context disrupted retention after all delays. In Experiment 1B, the failure of a contextual change to impair retention after 1 day was replicated. In Experiment 2, three same/different cue/context combinations were used as reminders in a reactivation paradigm, and all infants were tested 1 day later with their original training combination. A change in either the context or the cue significantly impaired the effectiveness of the reminder. These results reveal not only that contextual information is incorporated into the memory representations of very immature infants but also that memory retrieval is highly specific to the context in which an event was originally encoded. This specificity buffers against generalized memory retrieval after long retention intervals. The data are consistent with Reeves and Sperling's 1986 model of attention-gating. The context appears to serve as the initial gate for attention to potentially effective retrieval cues.