Memory constraints on infants' cross-situational statistical learning.
Memory constraints on infants' cross-situational statistical learning.
复制标题
婴儿跨情境统计学习的记忆限制。
DOI：
10.1016/j.cognition.2013.02.015
复制
发表时间：
2013
期刊：
影响因子：
3.4
通讯作者：
Johnson,ScottP
中科院分区：
文献类型：
--
作者：
Vlach,HaleyA;Johnson,ScottP
婴儿能够通过跟踪学习事件的共现概率，将语言标签映射到世界上的指称物，这种行为通常被称为跨情境统计学习。这项研究建立在现有研究的基础上，通过检查婴儿随着时间的推移聚集和检索单词相关配对的发展能力。16个月和20个月大的婴儿（N=32）接受了一项跨情境统计学习任务，其中一半的对象-标签配对是连续呈现的（集中呈现），另一半是跨时间分布的（交错呈现）。结果显示，幼儿在单词映射表现方面存在显著的发展差异；两个年龄组的婴儿都能够学习连续出现的配对，但只有20个月大的婴儿能够正确地推断出随时间分布的配对。这项研究揭示了婴儿跨越时间聚合和检索对象标签配对能力的重大限制，并挑战了基于检索过程成功和自动的跨情境统计学习理论。
Infants are able to map linguistic labels to referents in the world by tracking co-occurrence probabilities across learning events, a behavior often termed cross-situational statistical learning. This study builds upon existing research by examining infants’ developing ability to aggregate and retrieve word-referent pairings over time. 16- and 20-month-old infants (N=32) were presented with a cross-situational statistical learning task in which half of the object-label pairings were presented in immediate succession (massed) and half were distributed across time (interleaved). Results revealed striking developmental differences in word mapping performance; infants in both age groups were able to learn pairings presented in immediate succession, but only 20-month-old infants were able to correctly infer pairings distributed over time. This work reveals significant constraints on infants’ ability to aggregate and retrieve object-label pairings across time and challenges theories of cross-situational statistical learning that rest on retrieval processes as successful and automatic.