Development of short-term memory: Chunking in infancy

描述(申请人提供)：这项拟议的研究旨在研究组块的发展起源，组块是一种重组短期记忆表征以获得更大存储容量的过程。先前的研究发现，成年人和婴儿都被限制在短期记忆中存储大约3个项目。例如，婴儿可以记住最多3个隐藏的物体，但不能记住更大的数字。成年人可以通过将项目分组为有意义的单元或块来克服这一3项记忆限制。这样做可以极大地增加短期记忆容量。最近，我的实验室证明了这种能力存在于发育的早期：未受过训练的14个月大的婴儿也可以大块地吃东西(Feigenson&Halberda，2004)。当展示隐藏的物体数组时，如果物体最初被呈现在两个空间定义的2的集合中，婴儿记住的物体比没有这样的分组线索时更多，而这些物体看起来是一个4的集合。因此，婴儿可以使用空间线索将项目分块成集合，从而增加他们能记住的项目的总数。关于这种基本内存计算的性质，出现了几个重要的问题。首先，婴儿可以使用其他非空间线索将单个物品组合成一组吗？颜色或大小等知觉特征也是成人格式塔分组的强烈线索。目前尚不清楚这些知觉线索是否也能支持婴儿的组块。第二，婴儿可以根据语义知识组块吗？由先前获得的关于物体种类的知识驱动的这种内生计算的证据，将使婴儿和成人的块更紧密地结合在一起。第三，婴儿组块能力的参数限制是什么？虽然婴儿可以通过组块记住最多4个隐藏的物体，但目前还不清楚它们是否还能代表更多的物体。具体来说，是打开多少块，以及每个块中有多少个单独的物品，婴儿可以存储。综上所述，这项拟议的项目旨在阐明一种必要的记忆计算的起源。这些实验将揭示婴儿短期记忆的限制以及克服这些限制的机制与成年人的相似和不同之处。最终，建议的研究将有助于更广泛的目标，了解记忆发展的连续性和间断性，并可能对研究记忆计算障碍或改善记忆的策略产生影响。这项拟议的项目旨在通过研究正常发育的婴儿群体的短期记忆，为理解记忆发展这一更大的目标做出贡献。在记录诸如组块等记忆计算的早期出现时，该项目可能会对理解记忆障碍、阅读或依赖短期记忆的数学等技能方面的困难，或改善记忆的策略产生影响。

DESCRIPTION (provided by applicant): The proposed research aims to study the developmental origins of chunking, the process by which short-term memory representations are reorganized to enable greater storage capacity. Previous research has found that both adults and infants are limited to storing approximately 3 items in short-term memory. For example, infants can remember up to 3 hidden objects but fail to remember larger numbers. Adults can overcome this 3-item memory limit by grouping items into meaningful units or chunks. Doing so can vastly increase short-term memory capacity. Recently, my lab demonstrated that this ability is present early in development: untrained 14-month old infants can also chunk items (Feigenson & Halberda, 2004). When shown object arrays that were then hidden, infants remembered more objects if they were initially presented in two spatially-defined sets of 2 than when no such grouping cues were present and the objects appeared as a single set of 4. Hence, infants can use spatial cues to chunk items into sets, thereby increasing the total number of items they can remember. Several important questions arise regarding the nature of this fundamental memory computation. First, can infants use other, non-spatial cues to chunk individual items into sets? Perceptual features such as color or size are also strong Gestalt grouping cues for adults. It is unknown whether these perceptual cues can also support chunking in infants. Second, can infants chunk items based on semantic knowledge? Evidence for such an endogenous computation, driven by previously-acquired knowledge about object kinds, would more closely align infant and adult chunking. Third, what are the parametric limits on infants' chunking abilities? While infants can remember up to 4 hidden objects via chunking, it is unclear whether they can also represent more items. Specifically, is open how many chunks, and how many individual items within each chunk, infants can store. Taken together, the proposed project aims to elucidate the origins of an essential memory computation. The experiments will reveal the ways in which the limits on infants' short-term memory, and mechanisms that can allow those limits to be overcome, are similar to and different from those of mature adults. Ultimately, the proposed studies will contribute to the broader goal of understanding continuities and discontinuities in memory development, and may have repercussions for studying disorders of memory computations or for strategies for improving memory. The proposed project aims to contribute to the larger goal of understanding memory development by investigating short-term memory in a normally-developing population of infants. In documenting the early appearance of memory computations such as chunking, the project may have repercussions for understanding disorders of memory, difficulties in skills such as reading or math that rely on short-term memory, or strategies for improving memory.