Abstract
Working memory is a limited capacity system in which both infants and adults show an abrupt upper limit on the number of items they can store (Luck & Vogel 1997, Feigenson & Carey 2003, 2005). However, infants, unlike adults, show a catastrophic memory failure when capacity is exceeded. For example, when infants see 4 identical objects hidden and are allowed to retrieve 1 or 2 of them, they stop searching for the remaining objects (but search appropriately when only 1, 2, or 3 objects are hidden; Feigenson & Carey 2003, 2005). This suggests that infants try, but fail, to remember the entire array rather than storing just a manageable subset of these items.
Here we tested the effect of array heterogeneity on infants' memory capacity. In contrast to infants' previous failures to remember 4 identical objects, we find that 12-month-olds succeed with 4-object arrays that contain perceptually unique objects. Thus, array heterogeneity appears to improve infants' working memory. Two possibilities could account for this effect. First, heterogeneous arrays might allow infants to increase their working memory capacity to 4 items. Alternatively, infants might be able to store up to 3 heterogeneous items, without actually exceeding the 3-item limit of working memory (thus remembering “4” as “3”). In other words, heterogeneity might prevent the “catastrophic forgetting” typically observed when infants” working memory limit is exceeded. To decide between these possibilities, we presented infants with arrays of 4 heterogeneous objects and either allowed them to retrieve 2 or 3 of them before measuring any subsequent searching for the “missing” objects. Infants searched after retrieving 2 of 4, but not 3 of 4, unique objects, suggesting that heterogeneity allows infants to successfully remember a subset of items when presented with arrays exceeding their memory limits. Thus, heterogeneity appears to increase memory efficiency without actually expanding capacity.