DISTRIBUTED MEMORY AND THE REPRESENTATION OF GENERAL AND SPECIFIC INFORMATION
DISTRIBUTED MEMORY AND THE REPRESENTATION OF GENERAL AND SPECIFIC INFORMATION
复制标题
DOI：
10.1037/0096-3445.114.2.159
复制
发表时间：
1985-01-01
影响因子：
4.1
通讯作者：
RUMELHART, DE
中科院分区：
文献类型：
--
作者：
MCCLELLAND, JL;RUMELHART, DE
本章描述了信息处理和记忆的分布式模型，并将其应用于一般和特定信息的表示。该模型由大量简单的处理元件组成，这些元件通过可修改的连接相互发送兴奋和抑制信号。一般来说，分布式模型似乎为各种不同类型的模型提供了替代方案，这些模型假定抽象的，概括的表示，如原型，logogens，语义记忆表示，甚至语言规则。本章考虑了另一种概念化：一种分布式的，叠加的记忆方法。它的目的是表明，分布式模型提供了一种方法来解决抽象表示的具体困境。在分布式模型中，轨迹的叠加会自动导致抽象，尽管它仍然可以在一定程度上保留特定事件和经验的特质，或者事件和经验的特定重复子类。
This chapter describes a distributed model of information processing and memory and apply it to the representation of general and specific information. The model consists of a large number of simple processing elements which send excitatory and inhibitory signals to each other via modifiable connections. In general distributed models appear to provide alternatives to a variety of different kinds of models that postulate abstract, summary representations such as prototypes, logogens, semantic memory representations, or even linguistic rules. The chapter considers an alternative conceptualization: a distributed, superpositional approach to memory. It aims to show that distributed models provide a way to resolve the abstraction-representation of specifics dilemma. With a distributed model, the superposition of traces automatically results in abstraction though it can still preserve to some extent the idiosyncrasies of specific events and experiences, or of specific recurring subclasses of events and experiences.