Abstract
How do we retain accurate visual memories over a long time? Studies have demonstrated that successfully retrieving a memory increases the likelihood that it can be retrieved later. However, other studies shown that information provided during retrieval can alter how the original memory is reported (i.e., misinformation effect). These seemingly-contradictory findings suggest that retrieval calls the memory into a malleable state where it is augmented or modified by information available at that time. To test this, we first had participants encode 240 pictures of colored real objects. Then, memory for those objects was tested in two types of retrieval tasks (i.e., the recognition bias task and the baseline recall task) on the same day and the day after. In the recognition bias task, participants were first presented with a grayscale object image and indicated whether or not they remembered encoding its colored version. Subsequently, participants completed a two-second-long recognition practice in which they saw two colored versions of the same object and identified the one more similar to the encoded version. Participants then recalled the encoded object’s color. The baseline recall task was identical, except for the recognition practice, which was replaced by a two-second blank retention interval. We found that irrespective of retrieval type, memories retrieved on Day 1 were more likely to be retrieved on Day 2 than memories not tested on Day 1. Additionally, recall following recognition practice was biased towards the probe judged to be more similar to the encoded object. Interestingly, however, this recognition-induced memory bias was transient and did not influence memory recall on Day 2. Taken together, these data support our hypothesis that retrieval brings visual memories into a malleable state to be augmented or altered by memory-relevant information. Fortunately, recognition-induced memory biases may not permanently change the encoded representation.